Update UAV specifications and enhance performance metrics in the GPS-Denied system documentation. Refine acceptance criteria and clarify operational constraints for improved understanding.

2026-04-22 09:26:38 +00:00 · 2026-03-17 18:35:56 +02:00
parent 09e9ad06d3
commit 3ab47526bd
11 changed files with 1868 additions and 0 deletions
@@ -0,0 +1 @@
+I want to build a UAV plane for reconnaissance missions maximizing flight duration. Investigate what is the best frame material for that purpose
@@ -0,0 +1,562 @@
+# Solution Draft
+
+## Assessment Findings
+
+| Old Component Solution | Weak Point (functional/security/performance) | New Solution |
+|------------------------|----------------------------------------------|-------------|
+| ONNX Runtime as potential inference runtime for AI models | **Performance**: ONNX Runtime CUDA EP on Jetson Orin Nano is 7-8x slower than TRT standalone with default settings (tensor cores not utilized). Even TRT-EP shows up to 3x overhead on some models. | **Use native TRT Engine for all AI models**. Convert PyTorch → ONNX → trtexec → .engine. Load with tensorrt Python module. Eliminates ONNX Runtime dependency entirely. |
+| ONNX Runtime TRT-EP memory overhead | **Performance**: ONNX RT TRT-EP keeps serialized engine in memory (~420-440MB vs 130-140MB native TRT). Delta ~280-300MB PER MODEL. On 8GB shared memory, this wastes ~560-600MB for two models. | **Native TRT releases serialized blob after deserialization** → saves ~280-300MB per model. Total savings ~560-600MB — 7% of total memory. Critical given cuVSLAM map growth risk. |
+| No explicit TRT engine build step in offline pipeline | **Functional**: Draft03 mentions TRT FP16 but doesn't define the build workflow. When/where are engines built? | **Add TRT engine build to offline preparation pipeline**: After satellite tile download, run trtexec on Jetson to build .engine files. Store alongside tiles. One-time cost per model version. |
+| Cross-platform portability via ONNX Runtime | **Functional**: ONNX Runtime's primary value is cross-platform support. Our deployment is Jetson-only — this value is zero. We pay the performance/memory tax for unused portability. | **Drop ONNX Runtime**. Jetson Orin Nano Super is fixed deployment hardware. TRT Engine is the optimal runtime for NVIDIA-only deployment. |
+| No DLA offloading considered | **Performance**: Draft03 doesn't mention DLA. Jetson Orin Nano has NO DLA cores — only Orin NX (1-2) and AGX Orin (2) have DLA. | **Confirm: DLA offloading is NOT available on Orin Nano**. All inference must run on GPU (1024 CUDA cores, 16 tensor cores). This makes maximizing GPU efficiency via native TRT even more critical. |
+| LiteSAM MinGRU TRT compatibility risk | **Functional**: LiteSAM's subpixel refinement uses 4 stacked MinGRU layers over a 3×3 candidate window (seq_len=9). MinGRU gates depend only on input C_f (not h_{t-1}), so z_t/h̃_t are pre-computable. Ops are standard: Linear, Sigmoid, Mul, Add, ReLU, Tanh. Risk is LOW-MEDIUM — depends on whether implementation uses logcumsumexp (problematic) or simple loop (fine). Seq_len=9 makes this trivially rewritable. | **Day-one verification**: clone LiteSAM repo → torch.onnx.export → polygraphy inspect → trtexec --fp16. If export fails on MinGRU: rewrite forward() as unrolled loop (9 steps). **If LiteSAM cannot be made TRT-compatible: replace with EfficientLoFTR TRT** (proven TRT path via Coarse_LoFTR_TRT, 15.05M params, semi-dense matching). |
+| Camera shoots at ~3fps (draft03/04 hard constraint) | **Functional**: ADTI 20L V1 max continuous rate is **2.0 fps** (burst only, buffer-limited). ADTi recommends 1.5s per capture (**0.7 fps sustained**). 3fps is physically impossible. 2fps is not sustainable for multi-hour flights (buffer saturation + mechanical shutter wear). | **Revised to 0.7 fps sustained**. ADTI 20L V1 is the sole navigation camera — used for both cuVSLAM VO and satellite matching. At 70 km/h cruise and 600m altitude: 27.8m inter-frame displacement, ~175px pixel shift, **95.2% frame overlap** — within pyramid-assisted LK optical flow range. ESKF IMU prediction at 5-10Hz bridges 1.43s gaps between frames. Satellite matching triggered on keyframes from the same stream. Viewpro A40 Pro reserved for AI object detection only. |
+
+## UAV Platform
+
+### Airframe Configuration
+
+| Component | Specification | Weight |
+|-----------|--------------|--------|
+| Airframe | Custom 3.5m S-2 Glass Composite, Eppler 423 airfoil | ~4.5 kg |
+| Battery | 2x VANT Semi-Solid State 6S 30Ah (22.2V, 666Wh each) | 5.30 kg (2.65 kg each) |
+| Motor | T-Motor AT4125 KV540 (2000W peak, 5.5 kg thrust w/ APC 15x8) | 0.36 kg |
+| Propulsion Acc. | ESC, 15x8 Folding Propeller, Servos, Cables | ~0.50 kg |
+| Avionics | Pixhawk 6x + GPS | ~0.10 kg |
+| Computing | NVIDIA Jetson Orin Nano Super Dev Kit | ~0.30 kg |
+| Camera 1 | ADTI 20L V1 APS-C Camera + 16mm Lens | ~0.22 kg |
+| Camera 2 | Viewpro A40 Pro (A40TPro) AI Gimbal | ~0.85 kg |
+| Misc | Mounts, wiring, connectors | ~0.35 kg |
+| **Total AUW** | | **~12.5 kg** |
+
+T-Motor AT4125 KV540 is spec'd for 8-10 kg fixed-wing. At 12.5 kg AUW, static thrust-to-weight is ~0.44. Flyable but margins are tight — weight optimization should be monitored.
+
+### Flight Performance (Max Endurance)
+
+Assumptions: wingspan 3.5m, mean chord ~0.30m, wing area S ~1.05 m², AR ~11.7, Eppler 423 Cl_max ~1.8, cruise altitude 800-1000m (ρ ~1.10 kg/m³).
+
+| Parameter | Value |
+|-----------|-------|
+| Stall speed (900m altitude) | 10.6 m/s (38 km/h) |
+| Min-power speed (theoretical) | 12.0 m/s (43 km/h) — only 13% above stall, impractical |
+| **Max endurance cruise (1.3× stall margin)** | **14 m/s (50 km/h)** |
+| Best range speed | ~18 m/s (65 km/h) |
+
+| Energy Budget | Value |
+|---------------|-------|
+| Total battery energy | 1332 Wh (2 × 666 Wh) |
+| Usable (80% DoD) | 1066 Wh |
+| Climb to 900m (~5 min at 3 m/s) | −57 Wh |
+| 10% reserve | ×0.9 |
+| **Available for cruise** | **~908 Wh** |
+
+| Power Budget | Value |
+|--------------|-------|
+| Propulsion (L/D ~15, η_prop 0.65, η_motor 0.80) | ~212 W |
+| Electronics (Pixhawk + Jetson + cameras + gimbal + servos) | ~55 W |
+| **Total cruise power** | **~267 W** |
+
+| Endurance | Value |
+|-----------|-------|
+| **Max endurance (at 50 km/h)** | **~3.4 hours** |
+| Total mission (incl. climb + reserve) | ~3.5 hours |
+| Max range (at 65 km/h best-range speed) | ~209 km |
+
+### Camera 1: ADTI 20L V1 + 16mm Lens
+
+| Spec | Value |
+|------|-------|
+| Sensor | Sony CMOS APS-C, 23.2 × 15.4 mm |
+| Resolution | 5456 × 3632 (20 MP) |
+| Focal length | 16 mm |
+| Shutter | Mechanical global inter-mirror shutter (ADTI product line) |
+| Max continuous fps | **2.0 fps** (spec — burst rate, buffer-limited) |
+| Sustained capture rate | **~0.7 fps** (ADTi recommended 1.5s per capture) |
+| File formats | JPEG, RAW, RAW+JPEG |
+| HDMI video output | 1080p 24p/30p, 1440×1080 30p |
+| Weight | 118g body + ~100g lens |
+| ISP | Socionext Milbeaut |
+| Cooling | Active fan |
+
+**2.0 fps is a burst rate, not sustained.** The 2.0 fps spec is limited by the internal buffer (estimated 3-5 frames). Once the buffer fills, the camera throttles to the write pipeline speed of ~0.7 fps. The bottleneck chain: mechanical shutter actuation (~100-300ms) + ISP processing (demosaic, NR, JPEG compress) + storage write (~5-10 MB/frame JPEG). The 1.5s/capture recommendation guarantees the buffer never fills and accounts for thermal margin over multi-hour flights.
+
+**Mechanical shutter wear at sustained rates:**
+
+| Rate | Actuations per 3.5h flight | Est. flights before 150K shutter life | Est. flights before 500K shutter life |
+|------|---------------------------|---------------------------------------|---------------------------------------|
+| 2.0 fps (burst, unsustainable) | 25,200 | ~6 | ~20 |
+| 1.0 fps | 12,600 | ~12 | ~40 |
+| 0.7 fps (recommended) | 8,820 | ~17 | ~57 |
+
+The 20L V1 shutter lifespan is not documented. The higher-end 102PRO is rated at 500K actuations. The 20L as an entry-level model is likely 100K-150K. At 0.7 fps sustained, this gives ~11-57 flights depending on shutter rating.
+
+**Confirmed operational rate: 0.7 fps (JPEG mode) for both VO and satellite matching.**
+
+### Camera 1: Ground Coverage at Mission Altitude
+
+| Parameter | H = 600 m | H = 800 m | H = 1000 m |
+|-----------|-----------|-----------|------------|
+| Along-track footprint (15.4mm side) | 577 m | 770 m | 962 m |
+| Cross-track footprint (23.2mm side) | 870 m | 1160 m | 1450 m |
+| GSD | 15.9 cm/pixel | 21.3 cm/pixel | 26.6 cm/pixel |
+
+### Camera 1: Forward Overlap at 0.7 fps
+
+At 0.7 fps, distance between shots = V / 0.7.
+
+**At 70 km/h (19.4 m/s) — realistic cruise speed:**
+
+| Altitude | Along-track footprint | Shot gap (27.8m) | Forward overlap | Pixel shift |
+|----------|-----------------------|------------------|-----------------|-------------|
+| 600 m | 577 m | 27.8 m | **95.2%** | ~175 px |
+| 800 m | 770 m | 27.8 m | **96.4%** | ~131 px |
+| 1000 m | 962 m | 27.8 m | **97.1%** | ~105 px |
+
+**Across speed range (at 600m altitude, 0.7 fps):**
+
+| Speed | Frame gap | Pixel shift | Forward overlap |
+|-------|-----------|-------------|-----------------|
+| 50 km/h (14 m/s) | 20.0 m | ~126 px | 96.5% |
+| 70 km/h (19.4 m/s) | 27.8 m | ~175 px | 95.2% |
+| 90 km/h (25 m/s) | 35.7 m | ~224 px | 93.8% |
+
+Even at 90 km/h and the lowest altitude (600m), overlap remains >93%. The 16mm lens on APS-C at these altitudes produces a footprint so large that 0.7 fps provides massive redundancy for both VO and satellite matching.
+
+**For satellite matching specifically** (keyframe-based, every 5-10 camera frames):
+
+| Target overlap | Required gap (600m) | Time between shots (70 km/h) | Capture rate |
+|----------------|---------------------|------------------------------|--------------|
+| 80% | 115 m | 5.9 s | 0.17 fps |
+| 70% | 173 m | 8.9 s | 0.11 fps |
+| 60% | 231 m | 11.9 s | 0.084 fps |
+
+Even at the lowest altitude and highest speed, 1 satellite matching keyframe every 6-12 seconds gives 60-80% overlap.
+
+### Camera 2: Viewpro A40 Pro (A40TPro) AI Gimbal
+
+Dual EO/IR gimbal with AI tracking. Reserved for **AI object detection and tracking only** — not used for navigation. Operates independently from the navigation pipeline.
+
+### Camera Role Assignment
+
+| Role | Camera | Rate | Notes |
+|------|--------|------|-------|
+| Visual Odometry (cuVSLAM) | ADTI 20L V1 + 16mm | 0.7 fps (sustained) | Sole navigation camera. At 70 km/h: 27.8m/~175px displacement at 600m alt. 95%+ overlap. |
+| Satellite Image Matching | ADTI 20L V1 + 16mm | Keyframes from VO stream (~every 5-10 frames) | Same image stream as VO. Subset routed to satellite matcher on Stream B. |
+| AI Object Detection | Viewpro A40 Pro | Independent | Not part of navigation pipeline. |
+
+### cuVSLAM at 0.7 fps — Feasibility
+
+At 0.7 fps and 70 km/h cruise, inter-frame displacement is 27.8m. In pixel terms:
+
+| Altitude | Displacement | Pixel shift | % of image height | Overlap |
+|----------|-------------|-------------|-------------------|---------|
+| 600 m | 27.8 m | ~175 px | 4.8% | 95.2% |
+| 800 m | 27.8 m | ~131 px | 3.6% | 96.4% |
+| 1000 m | 27.8 m | ~105 px | 2.9% | 97.1% |
+
+cuVSLAM uses Lucas-Kanade optical flow with image pyramids. Standard LK handles 30-50px displacements on the base level; with 3-4 pyramid levels, effective search range extends to ~150-200px. At 600m altitude, the 175px shift is within this pyramid-assisted range. At 800-1000m, the shift drops to 105-131px — well within range.
+
+Key factors that make 0.7 fps viable at high altitude:
+- **Large footprint**: The 16mm lens on APS-C at 600-1000m produces 577-962m along-track coverage. The aircraft moves only 4-5% of the frame between shots.
+- **High texture from altitude**: At 600-1000m, each frame covers a large area with diverse terrain features (roads, field boundaries, structures) even in agricultural regions.
+- **IMU bridging**: cuVSLAM's built-in IMU integrator provides pose prediction during the 1.43s gap between frames. ESKF IMU prediction runs at 5-10Hz for continuous GPS_INPUT output.
+- **95%+ overlap**: Consecutive frames share >95% content — abundant features for matching.
+
+**Risk**: Over completely uniform terrain (e.g., single crop field filling entire 577m+ footprint), feature tracking may still fail. cuVSLAM falls back to IMU-only (~1s acceptable) then constant-velocity (~0.5s) before tracking loss. Satellite matching corrections every 5-10 frames bound accumulated drift.
+
+## Product Solution Description
+
+A real-time GPS-denied visual navigation system for fixed-wing UAVs, running on a Jetson Orin Nano Super (8GB). All AI model inference uses **native TensorRT Engine files** — no ONNX Runtime dependency. The system replaces the GPS module by sending MAVLink GPS_INPUT messages via pymavlink over UART at 5-10Hz.
+
+Position is determined by fusing: (1) CUDA-accelerated visual odometry (cuVSLAM — native CUDA) from ADTI 20L V1 at 0.7 fps sustained, (2) absolute position corrections from satellite image matching (LiteSAM or XFeat — TRT Engine FP16) using keyframes from the same ADTI image stream, and (3) IMU data from the flight controller via ESKF. Viewpro A40 Pro is reserved for AI object detection only.
+
+**Inference runtime decision**: Native TRT Engine over ONNX Runtime because:
+1. ONNX RT CUDA EP is 7-8x slower on Orin Nano (tensor core bug)
+2. ONNX RT TRT-EP wastes ~280-300MB per model (serialized engine retained in memory)
+3. Cross-platform portability has zero value — deployment is Jetson-only
+4. Native TRT provides direct CUDA stream control for pipelining with cuVSLAM
+
+**Hard constraint**: ADTI 20L V1 shoots at 0.7 fps sustained (1430ms interval). Full VO+ESKF pipeline within 400ms per frame. Satellite matching async on keyframes (every 5-10 camera frames). GPS_INPUT at 5-10Hz (ESKF IMU prediction fills gaps between camera frames).
+
+**AI Model Runtime Summary**:
+
+| Model | Runtime | Precision | Memory | Integration |
+|-------|---------|-----------|--------|-------------|
+| cuVSLAM | Native CUDA (PyCuVSLAM) | N/A (closed-source) | ~200-500MB | CUDA Stream A |
+| LiteSAM | TRT Engine | FP16 | ~50-80MB | CUDA Stream B |
+| XFeat | TRT Engine | FP16 | ~30-50MB | CUDA Stream B (fallback) |
+| ESKF | CPU (Python/C++) | FP64 | ~10MB | CPU thread |
+
+**Offline Preparation Pipeline** (before flight):
+1. Download satellite tiles → validate → pre-resize → store (existing)
+2. **NEW: Build TRT engines on Jetson** (one-time per model version)
+   - `trtexec --onnx=litesam_fp16.onnx --saveEngine=litesam.engine --fp16`
+   - `trtexec --onnx=xfeat.onnx --saveEngine=xfeat.engine --fp16`
+3. Copy tiles + engines to Jetson storage
+4. At startup: load engines + preload tiles into RAM
+
+```
+┌─────────────────────────────────────────────────────────────────────┐
+│                    OFFLINE (Before Flight)                           │
+│  1. Satellite Tiles → Download & Validate → Pre-resize → Store      │
+│     (Google Maps)     (≥0.5m/px, <2yr)     (matcher res)  (GeoHash)│
+│  2. TRT Engine Build (one-time per model version):                  │
+│     PyTorch model → reparameterize → ONNX export → trtexec --fp16  │
+│     Output: litesam.engine, xfeat.engine                            │
+│  3. Copy tiles + engines to Jetson storage                          │
+└─────────────────────────────────────────────────────────────────────┘
+                              │
+                              ▼
+┌─────────────────────────────────────────────────────────────────────┐
+│                    ONLINE (During Flight)                            │
+│                                                                     │
+│  STARTUP:                                                           │
+│  1. pymavlink → read GLOBAL_POSITION_INT → init ESKF                │
+│  2. Load TRT engines: litesam.engine + xfeat.engine                 │
+│     (tensorrt.Runtime → deserialize_cuda_engine → create_context)   │
+│  3. Allocate GPU buffers for TRT input/output (PyCUDA)              │
+│  4. Start cuVSLAM with ADTI 20L V1 camera stream                   │
+│  5. Preload satellite tiles ±2km into RAM                           │
+│  6. Begin GPS_INPUT output loop at 5-10Hz                           │
+│                                                                     │
+│  EVERY CAMERA FRAME (0.7fps sustained from ADTI 20L V1):           │
+│  ┌──────────────────────────────────────┐                           │
+│  │ ADTI 20L V1 → Downsample (CUDA)     │                           │
+│  │             → cuVSLAM VO+IMU (~9ms)  │ ← CUDA Stream A          │
+│  │             → ESKF measurement       │                           │
+│  └──────────────────────────────────────┘                           │
+│                                                                     │
+│  5-10Hz CONTINUOUS (IMU-driven between camera frames):              │
+│  ┌──────────────────────────────────────┐                           │
+│  │ ESKF IMU prediction → GPS_INPUT send │──→ Flight Controller      │
+│  └──────────────────────────────────────┘                           │
+│                                                                     │
+│  KEYFRAMES (every 5-10 camera frames, async):                       │
+│  ┌──────────────────────────────────────┐                           │
+│  │ Same ADTI frame → TRT inference (B): │                           │
+│  │   context.enqueue_v3(stream_B)       │──→ ESKF correction        │
+│  │   LiteSAM FP16 or XFeat FP16        │                           │
+│  └──────────────────────────────────────┘                           │
+│                                                                     │
+│  TELEMETRY (1Hz):                                                   │
+│  ┌──────────────────────────────────────┐                           │
+│  │ NAMED_VALUE_FLOAT: confidence, drift │──→ Ground Station         │
+│  └──────────────────────────────────────┘                           │
+└─────────────────────────────────────────────────────────────────────┘
+```
+
+## Architecture
+
+### Component: AI Model Inference Runtime
+
+| Solution | Tools | Advantages | Limitations | Performance | Memory | Fit |
+|----------|-------|-----------|-------------|------------|--------|-----|
+| Native TRT Engine | tensorrt Python + PyCUDA + trtexec | Optimal latency, minimal memory, full tensor core usage, direct CUDA stream control | Hardware-specific engines, manual buffer management, rebuild per TRT version | Optimal | ~50-130MB total (both models) | ✅ Best |
+| ONNX Runtime TRT-EP | onnxruntime + TensorRT EP | Auto-fallback for unsupported ops, simpler API, auto engine caching | +280-300MB per model, wrapper overhead, first-run latency spike | Near-parity (claimed), up to 3x slower (observed) | ~640-690MB total (both models) | ❌ Memory overhead unacceptable |
+| ONNX Runtime CUDA EP | onnxruntime + CUDA EP | Simplest API, broadest op support | 7-8x slower on Orin Nano (tensor core bug), no TRT optimizations | 7-8x slower | Standard | ❌ Performance unacceptable |
+| Torch-TensorRT | torch_tensorrt | AOT compilation, PyTorch-native, handles mixed TRT/PyTorch | Newer on Jetson, requires PyTorch runtime at inference | Near native TRT | PyTorch runtime ~500MB+ | ⚠️ Viable alternative if TRT export fails |
+
+**Selected**: **Native TRT Engine** — optimal performance and memory on our fixed NVIDIA hardware.
+
+**Fallback**: If any model has unsupported TRT ops (e.g., MinGRU in LiteSAM), use **Torch-TensorRT** for that specific model. Torch-TensorRT handles mixed TRT/PyTorch execution but requires PyTorch runtime in memory.
+
+### Component: TRT Engine Conversion Workflow
+
+**LiteSAM conversion**:
+1. Load PyTorch model with trained weights
+2. Reparameterize MobileOne backbone (collapse multi-branch → single Conv2d+BN)
+3. Export to ONNX: `torch.onnx.export(model, dummy_input, "litesam.onnx", opset_version=17)`
+4. Verify with polygraphy: `polygraphy inspect model litesam.onnx`
+5. Build engine on Jetson: `trtexec --onnx=litesam.onnx --saveEngine=litesam.engine --fp16 --memPoolSize=workspace:2048`
+6. Verify engine: `trtexec --loadEngine=litesam.engine --fp16`
+
+**XFeat conversion**:
+1. Load PyTorch model
+2. Export to ONNX: `torch.onnx.export(model, dummy_input, "xfeat.onnx", opset_version=17)`
+3. Build engine on Jetson: `trtexec --onnx=xfeat.onnx --saveEngine=xfeat.engine --fp16`
+4. Alternative: use XFeatTensorRT C++ implementation directly
+
+**INT8 quantization strategy** (optional, future optimization):
+- MobileOne backbone (CNN): INT8 safe with calibration data
+- TAIFormer (transformer attention): FP16 only — INT8 degrades accuracy
+- XFeat: evaluate INT8 on actual UAV-satellite pairs before deploying
+- Use nvidia-modelopt for calibration: `from modelopt.onnx.quantization import quantize`
+
+### Component: TRT Python Inference Wrapper
+
+Minimal wrapper class for TRT engine inference:
+
+```python
+import tensorrt as trt
+import pycuda.driver as cuda
+
+class TRTInference:
+    def __init__(self, engine_path, stream):
+        self.logger = trt.Logger(trt.Logger.WARNING)
+        self.runtime = trt.Runtime(self.logger)
+        with open(engine_path, 'rb') as f:
+            self.engine = self.runtime.deserialize_cuda_engine(f.read())
+        self.context = self.engine.create_execution_context()
+        self.stream = stream
+        self._allocate_buffers()
+
+    def _allocate_buffers(self):
+        self.inputs = {}
+        self.outputs = {}
+        for i in range(self.engine.num_io_tensors):
+            name = self.engine.get_tensor_name(i)
+            shape = self.engine.get_tensor_shape(name)
+            dtype = trt.nptype(self.engine.get_tensor_dtype(name))
+            size = trt.volume(shape)
+            device_mem = cuda.mem_alloc(size * np.dtype(dtype).itemsize)
+            self.context.set_tensor_address(name, int(device_mem))
+            mode = self.engine.get_tensor_mode(name)
+            if mode == trt.TensorIOMode.INPUT:
+                self.inputs[name] = (device_mem, shape, dtype)
+            else:
+                self.outputs[name] = (device_mem, shape, dtype)
+
+    def infer_async(self, input_data):
+        for name, data in input_data.items():
+            cuda.memcpy_htod_async(self.inputs[name][0], data, self.stream)
+        self.context.enqueue_v3(self.stream.handle)
+
+    def get_output(self):
+        results = {}
+        for name, (dev_mem, shape, dtype) in self.outputs.items():
+            host_mem = np.empty(shape, dtype=dtype)
+            cuda.memcpy_dtoh_async(host_mem, dev_mem, self.stream)
+        self.stream.synchronize()
+        return results
+```
+
+Key design: `infer_async()` + `get_output()` split enables pipelining with cuVSLAM on Stream A while satellite matching runs on Stream B.
+
+### Component: Visual Odometry (UPDATED — camera rate corrected)
+
+cuVSLAM — native CUDA library. Fed by **ADTI 20L V1 at 0.7 fps sustained** (previously assumed 3fps which exceeds camera hardware limit; 2.0 fps spec is burst-only, not sustainable). At 70 km/h cruise the inter-frame displacement is 27.8m — at 600m altitude this translates to ~175px (4.8% of frame), within pyramid-assisted LK optical flow range. At 800-1000m altitude the pixel shift drops to 105-131px. 95%+ frame overlap ensures abundant features for matching. ESKF IMU prediction at 5-10Hz fills the position output between sparse camera frames.
+
+### Component: Satellite Image Matching (UPDATED runtime + fallback chain)
+
+| Solution | Tools | Advantages | Limitations | Performance (est. Orin Nano Super TRT FP16) | Params | Fit |
+|----------|-------|-----------|-------------|----------------------------------------------|--------|-----|
+| LiteSAM (opt) TRT Engine FP16 @ 1280px | trtexec + tensorrt Python | Best satellite-aerial accuracy (RMSE@30=17.86m UAV-VisLoc), 6.31M params, smallest model | MinGRU TRT export needs verification (LOW-MEDIUM risk) | Est. ~165-330ms | 6.31M | ✅ Primary (if TRT export succeeds AND ≤200ms) |
+| EfficientLoFTR TRT Engine FP16 | trtexec + tensorrt Python | Proven TRT path (Coarse_LoFTR_TRT repo, 138 stars). Semi-dense. CVPR 2024. High accuracy. | 2.4x more params than LiteSAM. Requires einsum→elementary ops rewrite for TRT (documented in Coarse_LoFTR_TRT paper). | Est. ~200-400ms | 15.05M | ✅ Fallback if LiteSAM TRT fails |
+| XFeat TRT Engine FP16 | trtexec + tensorrt Python (or XFeatTensorRT C++) | Fastest. Proven TRT implementation. Lightweight. | General-purpose, not designed for cross-view satellite-aerial gap (but nadir-nadir gap is small). | Est. ~50-100ms | <5M | ✅ Speed fallback |
+
+**Decision tree (day-one on Orin Nano Super)**:
+1. Clone LiteSAM repo → reparameterize MobileOne → `torch.onnx.export()` → `polygraphy inspect`
+2. If ONNX export succeeds → `trtexec --onnx=litesam.onnx --saveEngine=litesam.engine --fp16`
+3. If MinGRU causes ONNX/TRT failure → rewrite MinGRU forward() as unrolled 9-step loop → retry
+4. If rewrite fails or accuracy degrades → **switch to EfficientLoFTR TRT**:
+   - Apply Coarse_LoFTR_TRT TRT-adaptation techniques (einsum replacement, etc.)
+   - Export to ONNX → trtexec --fp16
+   - Benchmark at 640×480 and 1280px
+5. Benchmark winner: **if ≤200ms → use it. If >200ms but ≤300ms → acceptable (async on Stream B). If >300ms → use XFeat TRT**
+
+**EfficientLoFTR TRT adaptation** (from Coarse_LoFTR_TRT paper, proven workflow):
+- Replace `torch.einsum()` with elementary ops (view, bmm, reshape, sum)
+- Replace any TRT-incompatible high-level PyTorch functions
+- Use ONNX export path (less memory required than Torch-TensorRT on 8GB device)
+- Knowledge distillation available for further parameter reduction if needed
+
+**Satellite matching cadence**: Keyframes selected from the ADTI VO stream every 5-10 frames (~every 2.5-14s depending on camera fps setting). At 800-1000m altitude and 14 m/s cruise, this yields 60-97% forward overlap between satellite match frames. Matching runs async on Stream B — does not block VO on Stream A.
+
+### Component: Sensor Fusion (UNCHANGED)
+ESKF — CPU-based mathematical filter, not affected.
+
+### Component: Flight Controller Integration (UNCHANGED)
+pymavlink — not affected by TRT migration.
+
+### Component: Ground Station Telemetry (UNCHANGED)
+MAVLink NAMED_VALUE_FLOAT — not affected.
+
+### Component: Startup & Lifecycle (UPDATED)
+
+**Updated startup sequence**:
+1. Boot Jetson → start GPS-Denied service (systemd)
+2. Connect to flight controller via pymavlink on UART
+3. Wait for heartbeat from flight controller
+4. **Initialize PyCUDA context**
+5. **Load TRT engines**: litesam.engine + xfeat.engine via tensorrt.Runtime.deserialize_cuda_engine()
+6. **Allocate GPU I/O buffers** for both models
+7. **Create CUDA streams**: Stream A (cuVSLAM), Stream B (satellite matching)
+8. Read GLOBAL_POSITION_INT → init ESKF
+9. Start cuVSLAM with ADTI 20L V1 camera frames
+10. Begin GPS_INPUT output loop at 5-10Hz
+11. Preload satellite tiles within ±2km into RAM
+12. System ready
+
+**Engine load time**: ~1-3 seconds per engine (deserialization from .engine file). One-time cost at startup.
+
+### Component: Thermal Management (UNCHANGED)
+Same adaptive pipeline. TRT engines are slightly more power-efficient than ONNX Runtime, but the difference is within noise.
+
+### Component: Object Localization (UNCHANGED)
+Not affected — trigonometric calculation, no AI inference.
+
+## Speed Optimization Techniques
+
+### 1. cuVSLAM for Visual Odometry (~9ms/frame)
+Fed by ADTI 20L V1 at 0.7 fps sustained. At 70 km/h cruise and 600m altitude, inter-frame displacement is 27.8m (~175px, 4.8% of frame). With pyramid-based LK optical flow (3-4 levels), effective search range is ~150-200px — 175px is within range. At 800-1000m altitude, pixel shift drops to 105-131px. 95%+ overlap between consecutive frames.
+
+### 2. Native TRT Engine Inference (NEW)
+All AI models run as pre-compiled TRT FP16 engines:
+- Engine files built offline with trtexec (one-time per model version)
+- Loaded at startup (~1-3s per engine)
+- Inference via context.enqueue_v3() on dedicated CUDA Stream B
+- GPU buffers pre-allocated — zero runtime allocation during flight
+- No ONNX Runtime dependency — no framework overhead
+
+Memory advantage over ONNX Runtime TRT-EP: ~560-600MB saved (both models combined).
+Latency advantage: eliminates ONNX wrapper overhead, guaranteed tensor core utilization.
+
+### 3. CUDA Stream Pipelining (REFINED)
+- Stream A: cuVSLAM VO from ADTI 20L V1 (~9ms) + ESKF fusion (~1ms)
+- Stream B: TRT engine inference for satellite matching (LiteSAM or XFeat, async, triggered on keyframe from same ADTI stream)
+- CPU: GPS_INPUT output loop, NAMED_VALUE_FLOAT, command listener, tile management
+- **NEW**: Both cuVSLAM and TRT engines use CUDA streams natively — no framework abstraction layer. Direct GPU scheduling.
+
+### 4-7. (UNCHANGED from draft03)
+Keyframe-based satellite matching, TensorRT FP16 optimization, proactive tile loading, 5-10Hz GPS_INPUT output — all unchanged.
+
+## Processing Time Budget
+
+### VO Frame (every ~1430ms from ADTI 20L V1 at 0.7 fps)
+
+| Step | Time | Notes |
+|------|------|-------|
+| ADTI image transfer | ~5-10ms | Trigger + readout |
+| Downsample (CUDA) | ~2ms | To cuVSLAM input resolution |
+| cuVSLAM VO+IMU | ~9ms | CUDA Stream A |
+| ESKF measurement update | ~1ms | CPU |
+| **Total** | **~17-22ms** | Well within 1430ms budget |
+
+Between camera frames, ESKF IMU prediction runs at 5-10Hz to maintain continuous GPS_INPUT output. The ~1.4s gap between frames is bridged entirely by IMU integration.
+
+### Keyframe Satellite Matching (every 5-10 camera frames, async CUDA Stream B)
+
+**Path A — LiteSAM TRT Engine FP16 at 1280px**:
+
+| Step | Time | Notes |
+|------|------|-------|
+| Image already in GPU (from VO) | ~0ms | Same frame used for VO and matching |
+| Load satellite tile | ~1ms | Pre-loaded in RAM |
+| Copy input to GPU buffer | <0.5ms | PyCUDA memcpy_htod_async |
+| LiteSAM TRT Engine FP16 | ≤200ms | context.enqueue_v3(stream_B) |
+| Copy output from GPU | <0.5ms | PyCUDA memcpy_dtoh_async |
+| Geometric pose (RANSAC) | ~5ms | Homography |
+| ESKF satellite update | ~1ms | Delayed measurement |
+| **Total** | **≤210ms** | Async on Stream B, does not block VO |
+
+**Path B — XFeat TRT Engine FP16**:
+
+| Step | Time | Notes |
+|------|------|-------|
+| XFeat TRT Engine inference | ~50-80ms | context.enqueue_v3(stream_B) |
+| Geometric verification (RANSAC) | ~5ms | |
+| ESKF satellite update | ~1ms | |
+| **Total** | **~60-90ms** | Async on Stream B |
+
+## Memory Budget (Jetson Orin Nano Super, 8GB shared)
+
+| Component | Memory (Native TRT) | Memory (ONNX RT TRT-EP) | Notes |
+|-----------|---------------------|--------------------------|-------|
+| OS + runtime | ~1.5GB | ~1.5GB | JetPack 6.2 + Python |
+| cuVSLAM | ~200-500MB | ~200-500MB | CUDA library + map |
+| **LiteSAM TRT engine** | **~50-80MB** | **~330-360MB** | Native TRT vs TRT-EP. If LiteSAM fails: EfficientLoFTR ~100-150MB |
+| **XFeat TRT engine** | **~30-50MB** | **~310-330MB** | Native TRT vs TRT-EP |
+| Preloaded satellite tiles | ~200MB | ~200MB | ±2km of flight plan |
+| pymavlink + MAVLink | ~20MB | ~20MB | |
+| FastAPI (local IPC) | ~50MB | ~50MB | |
+| ESKF + buffers | ~10MB | ~10MB | |
+| ONNX Runtime framework | **0MB** | **~150MB** | Eliminated with native TRT |
+| **Total** | **~2.1-2.9GB** | **~2.8-3.6GB** | |
+| **% of 8GB** | **26-36%** | **35-45%** | |
+| **Savings** | — | — | **~700MB saved with native TRT** |
+
+## Confidence Scoring → GPS_INPUT Mapping
+Unchanged from draft03.
+
+## Key Risks and Mitigations
+
+| Risk | Likelihood | Impact | Mitigation |
+|------|-----------|--------|------------|
+| **LiteSAM MinGRU ops unsupported in TRT 10.3** | LOW-MEDIUM | LiteSAM TRT export fails | Day-one verification: ONNX export → polygraphy → trtexec. If MinGRU fails: (1) rewrite as unrolled 9-step loop, (2) if still fails: **switch to EfficientLoFTR TRT** (proven TRT path, Coarse_LoFTR_TRT, 15.05M params). XFeat TRT as speed fallback. |
+| **TRT engine build OOM on 8GB Jetson** | LOW | Cannot build engines on target device | Our models are small (6.31M LiteSAM, <5M XFeat). OOM unlikely. If occurs: reduce --memPoolSize, or build on identical Orin Nano module with more headroom |
+| **Engine incompatibility after JetPack update** | MEDIUM | Must rebuild engines | Include engine rebuild in JetPack update procedure. Takes minutes per model. |
+| **MAVSDK cannot send GPS_INPUT** | CONFIRMED | Must use pymavlink | Unchanged from draft03 |
+| **cuVSLAM fails on low-texture terrain** | HIGH | Frequent tracking loss | ADTI at 0.7 fps means 27.8m inter-frame displacement at 70 km/h. At 600m+ altitude, pixel shift is 105-175px with 95%+ overlap — within pyramid-assisted LK range. HIGH risk remains over completely uniform terrain (single crop covering 577m+ footprint). IMU bridging + satellite matching corrections bound drift. |
+| **Thermal throttling** | MEDIUM | Satellite matching budget blown | Unchanged from draft03 |
+| LiteSAM TRT FP16 >200ms at 1280px | MEDIUM | Must use fallback matcher | Day-one benchmark. Fallback chain: EfficientLoFTR TRT (if ≤300ms) → XFeat TRT (if all >300ms) |
+| Google Maps satellite quality in conflict zone | HIGH | Satellite matching fails | Unchanged from draft03 |
+| **AUW exceeds AT4125 recommended range** | MEDIUM | Reduced endurance, motor thermal stress | 12.5 kg AUW vs 8-10 kg recommended. Monitor motor temps. Consider weight reduction (lighter gimbal, single battery for shorter missions). |
+| **cuVSLAM at 0.7 fps — inter-frame displacement** | MEDIUM | VO tracking loss on uniform terrain | At 0.7 fps and 70 km/h: ~175px displacement at 600m (4.8% of frame, 95.2% overlap). Within pyramid-assisted LK range (150-200px). At 800m+: drops to 105-131px. Mitigations: (1) cuVSLAM IMU integrator bridges 1.43s frame gaps, (2) ESKF IMU prediction at 5-10Hz fills position gaps, (3) satellite matching corrections every 5-10 frames bound drift. |
+| **ADTI mechanical shutter lifespan** | MEDIUM | Shutter replacement needed periodically | At 0.7 fps sustained over 3.5h flights: ~8,800 actuations/flight. Shutter life unknown for 20L (102PRO is 500K, entry-level likely 100-150K). Estimated 11-57 flights before replacement. Budget for shutter replacement as consumable. |
+
+## Testing Strategy
+
+### Integration / Functional Tests
+All tests from draft03 unchanged, plus:
+- **TRT engine load test**: Verify litesam.engine and xfeat.engine load successfully on Jetson Orin Nano Super
+- **TRT inference correctness**: Compare TRT engine output vs PyTorch reference output (max L1 error < 0.01)
+- **CUDA Stream B pipelining**: Verify satellite matching on Stream B does not block cuVSLAM on Stream A
+- **Engine pre-built validation**: Verify engine files from offline preparation work without rebuild at runtime
+- **ADTI 20L V1 sustained capture rate**: Verify camera sustains 0.7 fps in JPEG mode over extended periods (>30 min) without buffer overflow or overheating. Also test 1.0 fps to determine if higher sustained rate is achievable.
+- **ADTI trigger timing**: Verify camera trigger and image transfer pipeline delivers frames to cuVSLAM within acceptable latency (<50ms from trigger to GPU buffer)
+
+### Non-Functional Tests
+All tests from draft03 unchanged, plus:
+- **TRT engine build time**: Measure trtexec build time for LiteSAM and XFeat on Orin Nano Super (expected: 1-5 minutes each)
+- **TRT engine load time**: Measure deserialization time (expected: 1-3 seconds each)
+- **Memory comparison**: Measure actual GPU memory with native TRT vs ONNX RT TRT-EP for both models
+- **MinGRU TRT compatibility** (day-one blocker):
+  1. Clone LiteSAM repo, load pretrained weights
+  2. Reparameterize MobileOne backbone
+  3. `torch.onnx.export(model, dummy, "litesam.onnx", opset_version=17)`
+  4. `polygraphy inspect model litesam.onnx` — check for unsupported ops
+  5. `trtexec --onnx=litesam.onnx --saveEngine=litesam.engine --fp16`
+  6. If step 3 or 5 fails on MinGRU: rewrite MinGRU forward() as unrolled loop, retry
+  7. If still fails: switch to EfficientLoFTR, apply Coarse_LoFTR_TRT adaptation
+  8. Compare TRT output vs PyTorch reference (max L1 error < 0.01)
+- **EfficientLoFTR TRT fallback benchmark** (if LiteSAM fails): apply TRT adaptation from Coarse_LoFTR_TRT → ONNX → trtexec → measure latency at 640×480 and 1280px
+- **Tensor core utilization**: Verify with NSight that TRT engines use tensor cores (unlike ONNX RT CUDA EP)
+- **Flight endurance validation**: Ground-test full system power draw (propulsion + electronics) against 267W estimate. Verify ~3.4h endurance target.
+- **cuVSLAM at 0.7 fps**: Benchmark VO tracking quality, drift rate, and tracking loss frequency at 0.7 fps with ADTI 20L V1. Measure IMU integrator effectiveness for bridging 1.43s inter-frame gaps. Test at 600m and 800m altitude, over both textured and low-texture terrain.
+- **ADTI shutter durability**: Track shutter actuation count across flights. Monitor for shutter failure symptoms (missed frames, inconsistent exposure).
+
+## References
+- ONNX Runtime Issue #24085 (Jetson Orin Nano tensor core bug): https://github.com/microsoft/onnxruntime/issues/24085
+- ONNX Runtime Issue #20457 (TRT-EP memory overhead): https://github.com/microsoft/onnxruntime/issues/20457
+- ONNX Runtime Issue #12083 (TRT-EP vs native TRT): https://github.com/microsoft/onnxruntime/issues/12083
+- NVIDIA TensorRT 10 Python API: https://docs.nvidia.com/deeplearning/tensorrt/10.15.1/inference-library/python-api-docs.html
+- TensorRT Best Practices: https://docs.nvidia.com/deeplearning/tensorrt/latest/performance/best-practices.html
+- TensorRT engine hardware specificity: https://github.com/NVIDIA/TensorRT/issues/1920
+- trtexec ONNX conversion: https://nvidia-jetson.piveral.com/jetson-orin-nano/how-to-convert-onnx-to-engine-on-jetson-orin-nano-dev-board/
+- Torch-TensorRT JetPack 6.2: https://docs.pytorch.org/TensorRT/v2.10.0/getting_started/jetpack.html
+- XFeatTensorRT: https://github.com/PranavNedunghat/XFeatTensorRT
+- JetPack 6.2 Release Notes: https://docs.nvidia.com/jetson/archives/jetpack-archived/jetpack-62/release-notes/index.html
+- Jetson Orin Nano Super: https://developer.nvidia.com/blog/nvidia-jetson-orin-nano-developer-kit-gets-a-super-boost/
+- DLA on Jetson Orin: https://developer.nvidia.com/blog/maximizing-deep-learning-performance-on-nvidia-jetson-orin-with-dla/
+- EfficientLoFTR (CVPR 2024): https://github.com/zju3dv/EfficientLoFTR
+- EfficientLoFTR HuggingFace: https://huggingface.co/docs/transformers/en/model_doc/efficientloftr
+- Coarse_LoFTR_TRT (TRT for embedded): https://github.com/Kolkir/Coarse_LoFTR_TRT
+- Coarse_LoFTR_TRT paper: https://ar5iv.labs.arxiv.org/html/2202.00770
+- LoFTR_TRT: https://github.com/Kolkir/LoFTR_TRT
+- minGRU ("Were RNNs All We Needed?"): https://huggingface.co/papers/2410.01201
+- minGRU PyTorch implementation: https://github.com/lucidrains/minGRU-pytorch
+- LiteSAM paper (MinGRU details, Eqs 12-16): https://www.mdpi.com/2072-4292/17/19/3349
+- DALGlue (UAV feature matching, 2025): https://www.nature.com/articles/s41598-025-21602-5
+- ADTI 20L V1 specs: https://unmannedrc.com/products/adti-20l-v1-mapping-camera
+- ADTI 20L V1 user manual: https://docs.adti.camera/adti-20l-and-24l-v1-quick-start-guide/
+- T-Motor AT4125 KV540: https://uav-en.tmotor.com/2019/Motors_0429/247.html
+- VANT Semi-Solid State 6S 30Ah battery: https://www.xtbattery.com/370wh/kg-42v-high-energy-density-6s-12s-14s-18s-30ah-semi-solid-state-drone-battery/
+- All references from solution_draft03.md
+
+## Related Artifacts
+- AC Assessment: `_docs/00_research/gps_denied_nav/00_ac_assessment.md`
+- Research artifacts (this assessment): `_docs/00_research/trt_engine_migration/`
+- Previous research: `_docs/00_research/gps_denied_nav_v3/`
+- Tech stack evaluation: `_docs/01_solution/tech_stack.md`
+- Security analysis: `_docs/01_solution/security_analysis.md`
+- Previous draft: `_docs/01_solution/solution_draft04.md`
@@ -0,0 +1,41 @@
+# Acceptance Criteria Assessment
+
+## Acceptance Criteria
+
+| Criterion | Our Values | Researched Values | Cost/Timeline Impact | Status |
+|-----------|-----------|-------------------|---------------------|--------|
+| Flight duration | "Maximizing" (undefined) | 2-4 hours for electric fixed-wing in 5-15 kg MTOW class (Albatross: 4h, Vulture: 3.5-5h) | Higher endurance = larger battery + lighter frame = higher cost | Added — suggest target: ≥3 hours |
+| Payload capacity | 1.47 kg fixed | 1.47 kg is modest; benchmark platforms carry 4-8 kg in this class | Low payload relative to class = more weight budget for battery/fuel | Modified — no change needed, favorable constraint |
+| Frame weight | Not specified | Benchmark: 3.0-4.0 kg bare airframe for 3m wingspan class (Albatross: 3.35 kg) | Lighter frame = more battery weight = longer flight | Added — suggest target: ≤3.5 kg bare airframe |
+| MTOW | Not specified | 8-15 kg typical for this class | Drives wing sizing, motor selection, battery capacity | Added — suggest target: 8-12 kg |
+| Cruise speed | Not specified | 15-25 m/s typical for recon fixed-wing (Albatross: 19 m/s) | Slower cruise = longer endurance but less area coverage | Added — suggest target: 15-20 m/s |
+| Wingspan | Not specified | 2.5-3.5m for this MTOW class | Larger span = better L/D = longer endurance, but transport/handling harder | Added — suggest: 2.5-3.5m |
+| Battery energy density | Semi-solid state interest | 300-350 Wh/kg (semi-solid, 2025-2026 commercial products) vs 180-250 Wh/kg (LiPo) | Semi-solid ~2-3x cost of LiPo but 30% more flight time | Added — suggest: ≥300 Wh/kg (semi-solid) |
+| Budget | $100k total | Sufficient for custom composite airframe + avionics + batteries + ground station | $100k is generous for single prototype in this class | Modified — no change needed |
+| Operating temperature | Not specified | -20°C to 45°C is standard for commercial UAVs | Affects battery performance and material selection | Added — suggest: -10°C to 45°C |
+| Wind resistance | Not specified | 10-15 m/s sustained for fixed-wing recon | Affects structural requirements and endurance | Added — suggest: up to 12 m/s sustained |
+
+## Restrictions Assessment
+
+| Restriction | Our Values | Researched Values | Cost/Timeline Impact | Status |
+|-------------|-----------|-------------------|---------------------|--------|
+| Budget | $100k | Ample for 1 prototype. Carbon fiber airframe: $5-15k (custom tooling + manufacturing). Batteries: $2-5k. Avionics: $3-5k. Motor/prop/ESC: $1-2k. Ground station + comms: $5-10k. Integration + testing: $10-20k. | Well within range | Modified — no change needed |
+| Manufacturing access | None specified | Carbon fiber requires either outsourcing (CNC + layup vendors available globally) or basic workshop with vacuum bagging setup (~$2-5k investment) | Outsourcing is viable within budget; no blocker | Added — outsource recommended |
+| Regulatory | None specified | Sub-25 kg in most jurisdictions requires registration + remote pilot license. No specific material restrictions. | Minimal impact at this MTOW class | Added — follow local UAS regulations |
+| Payload (fixed) | 1.47 kg | Non-negotiable — mission equipment | No change | Added |
+| Frame material | Open investigation | Research strongly favors carbon fiber composite (CFRP) with foam-core sandwich construction | Drives the core research question | Added |
+
+## Key Findings
+
+1. **Flight duration target of ≥3 hours is realistic** for an electric fixed-wing in this class with semi-solid batteries. The Albatross achieves 4 hours with LiPo; semi-solid batteries would extend this further.
+
+2. **1.47 kg payload is light for this class** — leaves substantial weight budget for batteries, which directly translates to longer flight time. This is a favorable constraint.
+
+3. **Semi-solid state batteries (300-350 Wh/kg) are commercially available now** from multiple vendors (Grepow, Tattu, Herewin). They offer 30% more flight time than LiPo at 2-3x cost per Wh but with 4-6x cycle life, making TCO favorable.
+
+4. **$100k budget is generous** for a single prototype in this class. Typical custom composite UAV builds in this class cost $30-60k for first prototype including all subsystems.
+
+5. **Carbon fiber composite is the clear frontrunner** for frame material based on weight-to-stiffness ratio, which is the primary driver for endurance.
+
+## Sources
+- Source #1-#12 (see source registry)
@@ -0,0 +1,50 @@
+# Question Decomposition
+
+## Original Question
+"I want to build a UAV plane for reconnaissance missions maximizing flight duration. Investigate what is the best frame material for that purpose."
+
+## Active Mode
+Mode A (Initial Research) — no existing solution drafts found. Standalone mode.
+
+## Problem Context Summary
+- Fixed-wing UAV for reconnaissance missions
+- Primary objective: maximize flight duration
+- Payload: ~1.47 kg (ADTI 20L V1 camera 0.22 kg + Viewpro A40 Pro gimbal 0.85 kg + Jetson Orin Nano Super 0.30 kg + Pixhawk 6x + GPS 0.10 kg)
+- Battery: to be investigated, including semi-solid state options
+- Budget: ~$100k first iteration
+- No specific manufacturing method access yet
+- Standard fixed-wing UAV operating environment
+- No specific regulatory constraints stated
+
+## Classified Question Type
+**Decision Support** — the user needs to select the optimal frame material (and battery technology) to maximize a specific metric (flight duration) under budget and payload constraints.
+
+## Research Subject Boundary Definition
+| Dimension | Boundary |
+|-----------|----------|
+| Population | Electric fixed-wing UAVs in the 5-15 kg MTOW class |
+| Geography | Global — no regional restriction |
+| Timeframe | Current state-of-the-art (2024-2026) |
+| Level | Commercial/prosumer reconnaissance UAVs, not military large-scale platforms |
+
+## Decomposed Sub-Questions
+1. What frame materials are used in long-endurance fixed-wing UAVs and what are their properties (weight, strength, stiffness, cost, manufacturability)?
+2. How does frame material choice impact flight endurance for a ~1.5 kg payload fixed-wing UAV?
+3. What construction methods (monocoque, sandwich composite, foam-core) offer the best weight-to-strength for this class?
+4. What battery technologies (LiPo, Li-Ion, semi-solid state) are available for fixed-wing UAVs and what are their energy densities?
+5. What is the optimal airframe weight budget to maximize endurance given ~1.47 kg payload?
+6. What existing platforms in this class serve as benchmarks?
+7. What are realistic acceptance criteria for flight duration, MTOW, and cost?
+
+## Timeliness Sensitivity Assessment
+- **Research Topic**: UAV frame materials and semi-solid state batteries
+- **Sensitivity Level**: 🟠 High — battery technology (semi-solid state) is evolving rapidly; frame materials are more stable (🟡 Medium) but current commercial offerings matter
+- **Rationale**: Semi-solid state batteries are a fast-moving market with new products launching in 2025-2026. Frame materials are more established but new composite techniques are emerging.
+- **Source Time Window**: 12 months for battery tech, 2 years for frame materials
+- **Priority official sources to consult**:
+  1. Grepow, Tattu, Herewin official product pages (semi-solid batteries)
+  2. Applied Aeronautics, UAVMODEL product specs (benchmark platforms)
+  3. Academic papers on composite UAV structures (2023-2025)
+- **Key version information to verify**:
+  - Semi-solid battery energy density: currently 300-350 Wh/kg
+  - Tattu/Grepow product availability: confirmed commercial products in 2025-2026
@@ -0,0 +1,221 @@
+# Source Registry
+
+## Source #1
+- **Title**: Why Carbon Fiber Fixed Wing Drones Are the Future of Industrial UAVs — UAVMODEL
+- **Link**: https://www.uavmodel.com/blogs/news/skyeye-sr260-fixed-wing-drone-2600mm-long-endurance-mapping-amp-inspection
+- **Tier**: L2
+- **Publication Date**: 2025
+- **Timeliness Status**: ✅ Currently valid
+- **Target Audience**: Industrial/commercial UAV operators
+- **Research Boundary Match**: ✅ Full match
+- **Summary**: CFRP density 1.55-1.60 g/cm³ vs aluminum 2.7 g/cm³. Carbon fiber provides 40-50% weight reduction, superior vibration damping, thermal stability, and corrosion resistance for long-endurance fixed-wing drones.
+- **Related Sub-question**: 1, 2
+
+## Source #2
+- **Title**: SUX61 UAV Frame — Carbon Fiber, 8KG Payload, 91min Endurance
+- **Link**: https://aerojetparts.com/product/sux61-uav-frame-carbon-fiber-8kg-payload-91min-endurance/
+- **Tier**: L3
+- **Publication Date**: 2025
+- **Timeliness Status**: ✅ Currently valid
+- **Target Audience**: UAV builders/operators
+- **Research Boundary Match**: ✅ Full match
+- **Summary**: SUX61 carbon fiber frame: 3.4 kg airframe weight, 91-minute endurance, 8 kg payload. Uses 0.7mm thin-shell monocoque 3K carbon fiber via internal pressure molding.
+- **Related Sub-question**: 1, 6
+
+## Source #3
+- **Title**: Vanilla UAV 192-hour flight duration record — FAI
+- **Link**: https://www.fai.org/vanilla-uav-flight-duration-record
+- **Tier**: L1
+- **Publication Date**: 2021 (record event)
+- **Timeliness Status**: ✅ Currently valid (record still stands)
+- **Target Audience**: Aviation community
+- **Research Boundary Match**: ⚠️ Partial overlap — fuel-powered, much larger class, but demonstrates endurance design principles
+- **Summary**: Vanilla UAV set 192-hour 50-minute record. Demonstrates importance of systematic optimization across propulsion, avionics, and structural subsystems.
+- **Related Sub-question**: 6
+
+## Source #4
+- **Title**: Fluid Coupled Structural Analysis of EPS-Fiber-Reinforced Composite Wing — Springer
+- **Link**: https://link.springer.com/10.1007/s11029-024-10185-3
+- **Tier**: L1
+- **Publication Date**: 2024
+- **Timeliness Status**: ✅ Currently valid
+- **Target Audience**: Aerospace engineers
+- **Research Boundary Match**: ✅ Full match
+- **Summary**: EPS foam core + carbon/glass fiber composites achieved 30.5% wing weight reduction through topology optimization for MALE UAVs.
+- **Related Sub-question**: 3
+
+## Source #5
+- **Title**: Grepow Semi-Solid State Battery Product Page
+- **Link**: https://www.grepow.com/semi-solid-state-battery/300wh-kg-series-high-energy-density-battery-pack.html
+- **Tier**: L2
+- **Publication Date**: 2025
+- **Timeliness Status**: ✅ Currently valid
+- **Target Audience**: UAV manufacturers/integrators
+- **Research Boundary Match**: ✅ Full match
+- **Summary**: 300 Wh/kg series semi-solid state battery. NMC cathode, silicon-carbon anode, 2C charge, 3C continuous / 10C peak discharge, 1200+ cycles, -40°C to 60°C. 4S to 18S configurations.
+- **Related Sub-question**: 4
+
+## Source #6
+- **Title**: Tattu Semi-Solid State Battery for UAVs
+- **Link**: https://tattuworld.com/semi-solid-state-battery/
+- **Tier**: L2
+- **Publication Date**: 2025
+- **Timeliness Status**: ✅ Currently valid
+- **Target Audience**: Commercial drone operators
+- **Research Boundary Match**: ✅ Full match
+- **Summary**: 330-350 Wh/kg semi-solid batteries. Configurations from 1,550 mAh to 76,000 mAh, 11.4V to 68.4V. 500+ cycles at 90% retention. 30% flight endurance increase over LiPo.
+- **Related Sub-question**: 4
+
+## Source #7
+- **Title**: Herewin Semi-Solid State Battery Guide (2026 Update)
+- **Link**: https://www.herewinpower.com/blog/solid-state-drone-batteries-ultimate-guide/
+- **Tier**: L2
+- **Publication Date**: 2026
+- **Timeliness Status**: ✅ Currently valid
+- **Target Audience**: UAV manufacturers
+- **Research Boundary Match**: ✅ Full match
+- **Summary**: 300-400 Wh/kg at cell level, 303-313 Wh/kg at pack level. Silicon-carbon anodes (5-10% Si), high-Ni NCM cathode, 1000-3000 cycles. -20°C to 60°C operation.
+- **Related Sub-question**: 4
+
+## Source #8
+- **Title**: Applied Aeronautics Albatross UAV Specifications
+- **Link**: https://www.appliedaeronautics.com/albatross-uav
+- **Tier**: L2
+- **Publication Date**: 2024-2025
+- **Timeliness Status**: ✅ Currently valid
+- **Target Audience**: Commercial UAV operators
+- **Research Boundary Match**: ✅ Full match
+- **Summary**: Fiberglass+carbon fiber composite airframe. 3.35 kg bare airframe, 10 kg MTOW, up to 4 hours flight time, 4.5 kg payload capacity, 250+ km range.
+- **Related Sub-question**: 6, 1
+
+## Source #9
+- **Title**: Drone Frames: Carbon Fiber vs Aluminum — KingRaysCarbon
+- **Link**: https://kingrayscarbon.com/carbon-fiber-vs-aluminum-for-drone-frames-which-performs-better/
+- **Tier**: L3
+- **Publication Date**: 2024
+- **Timeliness Status**: ✅ Currently valid
+- **Target Audience**: UAV hobbyists and professionals
+- **Research Boundary Match**: ✅ Full match
+- **Summary**: Carbon fiber: tensile strength up to 3000 MPa, specific stiffness 113 vs aluminum 26. Carbon fiber is 40% lighter than aluminum. Fiberglass cheaper but heavier (2.46-2.58 g/cm³).
+- **Related Sub-question**: 1
+
+## Source #10
+- **Title**: Kevlar vs Carbon Fiber comparison — Dronecarbon
+- **Link**: https://www.dronecarbon.com/kevlar-vs-carbon-fiber_a9075.html
+- **Tier**: L3
+- **Publication Date**: 2024
+- **Timeliness Status**: ✅ Currently valid
+- **Target Audience**: UAV builders
+- **Research Boundary Match**: ✅ Full match
+- **Summary**: Kevlar: 5x stronger than steel by weight, superior impact absorption, lower cost than CF. But heavier than CF, poor compressive strength, UV/moisture sensitive, difficult to machine.
+- **Related Sub-question**: 1
+
+## Source #11
+- **Title**: LFP vs LiPo vs Semi-Solid Industrial Drone Batteries 2026 — Herewin
+- **Link**: https://www.herewinpower.com/blog/lfp-vs-lipo-vs-semi-solid-industrial-drone-batteries-2026-roi-safety-and-performance/
+- **Tier**: L2
+- **Publication Date**: 2026
+- **Timeliness Status**: ✅ Currently valid
+- **Target Audience**: UAV manufacturers/operators
+- **Research Boundary Match**: ✅ Full match
+- **Summary**: Comparison of LFP, LiPo, and semi-solid batteries for industrial drones. Semi-solid achieves highest energy density and best long-term ROI.
+- **Related Sub-question**: 4
+
+## Source #12
+- **Title**: ASTM F3563-22 — Standard Specification for Large Fixed-Wing UAS
+- **Link**: https://www.astm.org/f3563-22.html
+- **Tier**: L1
+- **Publication Date**: 2022
+- **Timeliness Status**: ✅ Currently valid
+- **Target Audience**: UAV manufacturers, CAAs
+- **Research Boundary Match**: ⚠️ Partial overlap — covers larger UAS but defines industry consensus standards
+- **Summary**: Industry consensus standard for design and construction of large fixed-wing UAS. Accepted by CAAs as Means of Compliance.
+- **Related Sub-question**: 7
+
+## Source #13
+- **Title**: DeltaQuad Evo Government Edition Specifications
+- **Link**: https://docs.deltaquad.com/gov/vehicle-specifications
+- **Tier**: L1
+- **Publication Date**: 2025
+- **Timeliness Status**: ✅ Currently valid
+- **Target Audience**: UAV operators/integrators
+- **Research Boundary Match**: ✅ Full match
+- **Summary**: Empty weight 4.8 kg, MTOW 10 kg, wingspan 269 cm. Uses semi-solid state Li-ion batteries (6S, 22 Ah). Airframe: fiberglass, carbon, Kevlar, composite. 4h32m endurance (dual battery), 8h55m record with solid-state batteries.
+- **Related Sub-question**: 6, 1, 4
+
+## Source #14
+- **Title**: T700 vs T800 Carbon Fiber — Practical Selection Guide
+- **Link**: https://www.carbonfibermaterial.com/t700-vs-t800-carbon-fiber-a-practical-guide-for-material-selection/
+- **Tier**: L3
+- **Publication Date**: 2024
+- **Timeliness Status**: ✅ Currently valid
+- **Target Audience**: Composite engineers
+- **Research Boundary Match**: ✅ Full match
+- **Summary**: T700: 4900 MPa tensile, 230 GPa modulus, ~$18/m². T800: 5880 MPa, 294 GPa, ~$26/m². T700 recommended for UAVs — better impact resistance, lower cost, nearly same density.
+- **Related Sub-question**: 1
+
+## Source #15
+- **Title**: CFRP Manufacturing Methods Comparison (VI vs VB vs HLU)
+- **Link**: https://ejournal.brin.go.id/ijoa/article/view/286
+- **Tier**: L1
+- **Publication Date**: 2024
+- **Timeliness Status**: ✅ Currently valid
+- **Target Audience**: Aerospace composite engineers
+- **Research Boundary Match**: ✅ Full match
+- **Summary**: Vacuum infusion: 71% higher compressive/shear strength than hand layup, 53% higher than vacuum bagging. Prepreg achieves <0.5% void content vs 2-5% wet layup.
+- **Related Sub-question**: 3
+
+## Source #16
+- **Title**: Rohacell vs Honeycomb, Balsa & PVC Foam — Chem-Craft
+- **Link**: https://chem-craft.com/blog/comparative-analysis-rohacell-vs-traditional-materials-in-composite-engineering/
+- **Tier**: L3
+- **Publication Date**: 2024
+- **Timeliness Status**: ✅ Currently valid
+- **Target Audience**: Composite engineers
+- **Research Boundary Match**: ✅ Full match
+- **Summary**: Rohacell PMI: highest stiffness-to-weight, closed-cell, withstands autoclave temps. XPS: good cost/performance middle ground. EPS: cheapest but lowest strength. PVC: moderate cost/performance.
+- **Related Sub-question**: 3
+
+## Source #17
+- **Title**: LFP vs LiPo vs Semi-Solid Industrial Drone Batteries 2026 — Herewin
+- **Link**: https://www.herewinpower.com/blog/lfp-vs-lipo-vs-semi-solid-industrial-drone-batteries-2026-roi-safety-and-performance/
+- **Tier**: L2
+- **Publication Date**: 2026
+- **Timeliness Status**: ✅ Currently valid
+- **Target Audience**: UAV manufacturers/operators
+- **Research Boundary Match**: ✅ Full match
+- **Summary**: Semi-solid 300-400 Wh/kg, 800-1200 cycles. LiPo 100-200 Wh/kg, 200-500 cycles. Li-ion 200-250 Wh/kg, 500-800 cycles. Semi-solid reduces internal temp rise by 60% vs LiPo.
+- **Related Sub-question**: 4
+
+## Source #18
+- **Title**: Carbon-Kevlar Hybrid Fabric Properties — Impact Materials
+- **Link**: https://ictmaterial.com/what-is-carbon-kevlar-hybrid-fabric-properties-and-use-cases/
+- **Tier**: L3
+- **Publication Date**: 2025
+- **Timeliness Status**: ✅ Currently valid
+- **Target Audience**: Composite engineers
+- **Research Boundary Match**: ✅ Full match
+- **Summary**: Carbon-Kevlar hybrid: 800-1200 MPa tensile, 70-90 GPa modulus, 25-40% lighter than aluminum. Superior crash survivability via Kevlar's energy absorption.
+- **Related Sub-question**: 1
+
+## Source #19
+- **Title**: Scabro Innovations — UAV Composite Prototyping
+- **Link**: https://scabroinnovations.com/diensten/composite-airframe-prototyping/
+- **Tier**: L3
+- **Publication Date**: 2025
+- **Timeliness Status**: ✅ Currently valid
+- **Target Audience**: UAV developers
+- **Research Boundary Match**: ✅ Full match
+- **Summary**: Full-service composite airframe prototyping. Flexible tooling from single prototype to production. Semi-assembled airframes, wings, full systems with wiring/sensor integration.
+- **Related Sub-question**: 3
+
+## Source #20
+- **Title**: Tattu 330Wh/Kg Semi-Solid 33000mAh 22.2V 6S Product Page
+- **Link**: https://www.tattuworld.com/semi-solid-state-battery/semi-solid-330wh-kg-33000mah-22-2v-10c-6s-battery.html
+- **Tier**: L2
+- **Publication Date**: 2025
+- **Timeliness Status**: ✅ Currently valid
+- **Target Audience**: UAV operators
+- **Research Boundary Match**: ✅ Full match
+- **Summary**: 330 Wh/kg, 33000 mAh, 22.2V (6S), 10C peak discharge, weight 2324g, 732.6 Wh energy per pack. Dimensions: 210x93x60.5mm.
+- **Related Sub-question**: 4
@@ -0,0 +1,161 @@
+# Fact Cards
+
+## Fact #1
+- **Statement**: Carbon fiber reinforced polymer (CFRP) has a density of 1.55-1.60 g/cm³, compared to aluminum at 2.7 g/cm³ and fiberglass at 2.46-2.58 g/cm³, resulting in 40-50% lighter frames for equivalent stiffness.
+- **Source**: Source #1, #9
+- **Phase**: Phase 1
+- **Target Audience**: All fixed-wing UAV classes
+- **Confidence**: ✅ High
+- **Related Dimension**: Weight / material density
+
+## Fact #2
+- **Statement**: CFRP tensile strength reaches up to 3000 MPa with specific stiffness of 113, compared to aluminum at 26 and titanium at 25.
+- **Source**: Source #9
+- **Phase**: Phase 1
+- **Target Audience**: All UAV classes
+- **Confidence**: ✅ High
+- **Related Dimension**: Structural strength
+
+## Fact #3
+- **Statement**: The Albatross UAV (fiberglass + carbon fiber composite) weighs 3.35 kg bare airframe, achieves 10 kg MTOW, and flies up to 4 hours with payload capacity of 4.5 kg.
+- **Source**: Source #8
+- **Phase**: Phase 1
+- **Target Audience**: Commercial fixed-wing UAVs in our target class
+- **Confidence**: ✅ High
+- **Related Dimension**: Benchmark platform
+
+## Fact #4
+- **Statement**: EPS foam core reinforced with carbon/glass fiber composites achieved 30.5% wing weight reduction through topology optimization for MALE UAVs.
+- **Source**: Source #4
+- **Phase**: Phase 1
+- **Target Audience**: Fixed-wing UAV designers
+- **Confidence**: ✅ High (peer-reviewed)
+- **Related Dimension**: Construction method
+
+## Fact #5
+- **Statement**: Semi-solid state batteries currently achieve 300-350 Wh/kg at cell level, with pack-level targets of 303-313 Wh/kg. This is 30-50% higher than traditional LiPo (150-250 Wh/kg).
+- **Source**: Source #5, #6, #7
+- **Phase**: Phase 1
+- **Target Audience**: UAV battery selection
+- **Confidence**: ✅ High (multiple manufacturer confirmations)
+- **Related Dimension**: Battery energy density
+
+## Fact #6
+- **Statement**: Tattu semi-solid batteries: 330-350 Wh/kg, capacities from 1,550-76,000 mAh, voltages 11.4-68.4V, 500+ cycles at 90% retention, 10C peak discharge.
+- **Source**: Source #6
+- **Phase**: Phase 1
+- **Target Audience**: Commercial UAV operators
+- **Confidence**: ✅ High (manufacturer spec)
+- **Related Dimension**: Battery specifications
+
+## Fact #7
+- **Statement**: Grepow semi-solid batteries: 300 Wh/kg series, 2C charge, 3C continuous / 10C peak discharge, 1200+ cycles, -40°C to 60°C, multiple S configurations (4S-18S).
+- **Source**: Source #5
+- **Phase**: Phase 1
+- **Target Audience**: Commercial UAV operators
+- **Confidence**: ✅ High (manufacturer spec)
+- **Related Dimension**: Battery specifications
+
+## Fact #8
+- **Statement**: Semi-solid state batteries deliver 800-1200+ cycle life vs 200-300 for traditional LiPo, retaining >80% capacity after 1000+ cycles.
+- **Source**: Source #5, #7
+- **Phase**: Phase 1
+- **Target Audience**: UAV operators
+- **Confidence**: ✅ High
+- **Related Dimension**: Battery longevity / cost of ownership
+
+## Fact #9
+- **Statement**: Kevlar is heavier than carbon fiber, has poor compressive strength, is UV/moisture sensitive, and difficult to machine — making it inferior to CFRP for UAV primary structure despite superior impact absorption.
+- **Source**: Source #10
+- **Phase**: Phase 1
+- **Target Audience**: UAV frame material selection
+- **Confidence**: ✅ High
+- **Related Dimension**: Material comparison
+
+## Fact #10
+- **Statement**: Carbon fiber-balsa sandwich structures provide excellent mechanical properties. Balsa core is ultra-lightweight but susceptible to moisture absorption. Modern approach favors EPS/Rohacell foam cores for moisture immunity.
+- **Source**: Source #4, Springer comparative analysis
+- **Phase**: Phase 1
+- **Target Audience**: UAV wing designers
+- **Confidence**: ✅ High
+- **Related Dimension**: Construction method
+
+## Fact #11
+- **Statement**: SUX61 carbon fiber frame achieves 91-minute endurance with 3.4 kg airframe weight and 8 kg payload using 0.7mm thin-shell monocoque 3K carbon fiber.
+- **Source**: Source #2
+- **Phase**: Phase 1
+- **Target Audience**: Fixed-wing UAV builders in our class
+- **Confidence**: ⚠️ Medium (single manufacturer claim)
+- **Related Dimension**: Benchmark platform
+
+## Fact #12
+- **Statement**: For electric propeller-driven aircraft, maximum endurance occurs at minimum power required speed (~76% of best-range speed). Endurance is directly proportional to battery energy and L/D ratio, inversely proportional to weight.
+- **Source**: FIRGELLI endurance calculator, general aerospace engineering
+- **Phase**: Phase 1
+- **Target Audience**: All electric fixed-wing UAVs
+- **Confidence**: ✅ High (established physics)
+- **Related Dimension**: Endurance optimization
+
+## Fact #13
+- **Statement**: Custom carbon fiber UAV airframes range from $20-50 for small CNC-cut frames to ~$3000 for large industrial frames. Full custom composite airframe development with tooling would be significantly more.
+- **Source**: Multiple manufacturer listings
+- **Phase**: Phase 1
+- **Target Audience**: UAV builders
+- **Confidence**: ⚠️ Medium (prices vary widely by specification)
+- **Related Dimension**: Cost
+
+## Fact #14
+- **Statement**: T700 carbon fiber: 4900 MPa tensile, 230 GPa modulus, ~$18/m². T800: 5880 MPa, 294 GPa, ~$26/m² (44% premium). T700 has better impact tolerance due to higher elongation at break. Density is nearly identical (1.80 vs 1.81 g/cm³).
+- **Source**: Source #14
+- **Phase**: Phase 2
+- **Target Audience**: UAV composite designers
+- **Confidence**: ✅ High
+- **Related Dimension**: Material grade selection
+
+## Fact #15
+- **Statement**: Vacuum infusion produces 71% higher compressive strength and 71% higher shear strength than hand layup; 53% higher than vacuum bagging. Prepreg achieves <0.5% void content vs 2-5% for wet layup.
+- **Source**: Source #15
+- **Phase**: Phase 2
+- **Target Audience**: Composite manufacturers
+- **Confidence**: ✅ High (peer-reviewed)
+- **Related Dimension**: Manufacturing method
+
+## Fact #16
+- **Statement**: DeltaQuad Evo: 4.8 kg empty, 10 kg MTOW, 269 cm wingspan. Uses fiberglass + carbon + Kevlar composite. Semi-solid state 6S 22Ah batteries. Achieved 8h55m endurance record with solid-state batteries. Standard endurance 4h32m with dual semi-solid batteries.
+- **Source**: Source #13
+- **Phase**: Phase 2
+- **Target Audience**: Fixed-wing UAV designers in our target class
+- **Confidence**: ✅ High (manufacturer + FAI-type record)
+- **Related Dimension**: Benchmark validation
+
+## Fact #17
+- **Statement**: Carbon-Kevlar hybrid fabric: 800-1200 MPa tensile, 70-90 GPa modulus, 25-40% lighter than aluminum. Superior crash survivability via Kevlar energy absorption. But Kevlar is UV-sensitive, moisture-absorbing, and very difficult to machine.
+- **Source**: Source #18
+- **Phase**: Phase 2
+- **Target Audience**: UAV structural designers
+- **Confidence**: ✅ High
+- **Related Dimension**: Hybrid material approach
+
+## Fact #18
+- **Statement**: PVC foam (Divinycell H-series): closed-cell, density 40-250 kg/m³, moisture-immune, handles 80°C cure. Rohacell PMI: highest stiffness/weight, 180°C+, but 3-5x more expensive. XPS: cheapest closed-cell option but limited to 75°C.
+- **Source**: Source #16
+- **Phase**: Phase 2
+- **Target Audience**: Composite wing designers
+- **Confidence**: ✅ High
+- **Related Dimension**: Foam core selection
+
+## Fact #19
+- **Statement**: For electric fixed-wing UAV, endurance = usable battery energy / total system power. Payload electronics (Jetson Orin Nano ~15-25W, camera+gimbal ~10-15W) add ~30W to cruise power, reducing endurance by ~15-20% compared to calculations ignoring payload power.
+- **Source**: General aerospace engineering + manufacturer specs
+- **Phase**: Phase 2
+- **Target Audience**: UAV system designers
+- **Confidence**: ✅ High
+- **Related Dimension**: Endurance calculation
+
+## Fact #20
+- **Statement**: Tattu 330Wh/kg 6S 33000mAh: weight 2324g, energy 732.6 Wh, dimensions 210×93×60.5mm. At pack level: 732.6/2.324 = 315 Wh/kg actual. 10C peak discharge, XT90-S connector.
+- **Source**: Source #20
+- **Phase**: Phase 2
+- **Target Audience**: UAV battery integration
+- **Confidence**: ✅ High (manufacturer spec)
+- **Related Dimension**: Battery sizing
@@ -0,0 +1,69 @@
+# Comparison Framework
+
+## Selected Framework Type
+Decision Support — selecting optimal frame material and battery to maximize flight endurance under budget/payload constraints.
+
+## Selected Dimensions
+1. Weight (density, specific weight)
+2. Structural performance (stiffness, strength, fatigue)
+3. Impact resistance / crash survivability
+4. Manufacturing complexity / accessibility
+5. Cost (material + manufacturing + tooling)
+6. Environmental durability (temperature, moisture, UV)
+7. Repairability in field conditions
+8. Endurance impact (calculated flight time contribution)
+
+## Component 1: Frame Material
+
+| Dimension | CFRP (T700) | Fiberglass (E-glass) | Carbon-Kevlar Hybrid | Aluminum 6061-T6 |
+|-----------|-------------|---------------------|----------------------|-------------------|
+| Density (g/cm³) | 1.55-1.60 | 2.46-2.58 | ~1.45 | 2.70 |
+| Tensile strength (MPa) | 3000-4900 | 800-1500 | 800-1200 | 310 |
+| Specific stiffness | 113 | ~28 | 48-62 | 26 |
+| Impact resistance | Low (brittle) | Medium | High (Kevlar absorption) | High (ductile) |
+| Cost (relative) | High ($18/m² T700) | Low (~$5/m²) | Very High (~$30/m²) | Low |
+| Manufacturability | Medium (requires curing) | Easy (room-temp cure) | Difficult (Kevlar hard to machine) | Easy (CNC) |
+| Moisture resistance | Excellent | Good | Good (Kevlar absorbs) | Excellent |
+| UV resistance | Good (with coating) | Good | Poor (Kevlar degrades) | Excellent |
+| Repairability | Difficult | Easy | Very difficult | Easy |
+| Weight savings vs Al | 40-50% | 5-10% | 45-55% | Baseline |
+| Factual Basis | Fact #1, #2, #9 | Fact #1, #3 | Source #18 | Fact #1 |
+
+## Component 2: Construction Method
+
+| Dimension | Sandwich (foam core + CF skin) | Monocoque (solid CF shell) | Spar + Rib + Skin (traditional) |
+|-----------|-------------------------------|---------------------------|-------------------------------|
+| Weight efficiency | Excellent (30% lighter) | Good | Moderate |
+| Stiffness per weight | Highest | High | Moderate |
+| Manufacturing complexity | Medium (requires core + layup) | Medium (requires mold) | Higher (many parts) |
+| Tooling cost | Medium | High (precise molds) | Low-Medium |
+| Repairability | Moderate | Difficult | Good (replace parts) |
+| Best for | Wings, fuselage panels | Fuselage, nacelles | Prototype/custom builds |
+| Factual Basis | Fact #4, #10 | Fact #11 | General aerospace |
+
+## Component 3: Battery Technology
+
+| Dimension | Semi-Solid State | Li-Ion (21700/18650) | LiPo |
+|-----------|-----------------|---------------------|------|
+| Energy density (Wh/kg) | 300-350 (pack: 310) | 200-250 | 150-200 |
+| Cycle life | 800-1200 | 500-800 | 200-500 |
+| Peak discharge (C) | 10C | 3-5C | 25-50C |
+| Continuous discharge (C) | 3-5C | 1-3C | 5-10C |
+| Operating temp | -20°C to 60°C | -20°C to 60°C | 0°C to 50°C |
+| Safety (thermal runaway) | Very low risk | Low risk | Medium risk |
+| Cost per Wh | ~$0.50-0.80 | ~$0.20-0.35 | ~$0.15-0.25 |
+| Availability | Commercial (Tattu, Grepow) | Widely available | Widely available |
+| Endurance impact (same weight) | Baseline (best) | -20 to -30% | -40 to -50% |
+| Factual Basis | Fact #5-8, Source #17, #20 | Source #17 | Source #17 |
+
+## Component 4: Foam Core Selection (for sandwich construction)
+
+| Dimension | Rohacell (PMI) | XPS | PVC (Divinycell) | EPS |
+|-----------|---------------|-----|-------------------|-----|
+| Density (kg/m³) | 32-110 | 25-45 | 40-250 | 15-30 |
+| Compressive strength | Highest | Moderate (200-500 kPa) | High | Lowest |
+| Temp resistance | High (180°C+) | Low (75°C) | Moderate (80°C) | Low (70°C) |
+| Moisture absorption | Very low | Low | Low | Medium |
+| Cost | Very high | Low | Medium | Very low |
+| Best for | High-performance wings | Budget wings | General-purpose | Prototypes only |
+| Factual Basis | Source #16 | Source #16 | Source #16 | Fact #4 |
@@ -0,0 +1,115 @@
+# Reasoning Chain
+
+## Dimension 1: Frame Material Selection
+
+### Fact Confirmation
+CFRP (T700) has density 1.55-1.60 g/cm³ with specific stiffness of 113 (Fact #1, #2). This is 40-50% lighter than aluminum for equivalent stiffness. Fiberglass at 2.46-2.58 g/cm³ offers only 5-10% weight savings over aluminum. The DeltaQuad Evo (Source #13) uses a hybrid of fiberglass, carbon, and Kevlar — achieving 4.8 kg empty weight at 269 cm wingspan.
+
+### Reference Comparison
+The Albatross UAV uses fiberglass + carbon fiber hybrid and achieves 3.35 kg bare airframe at similar wingspan (~3m). The DeltaQuad uses a tri-material hybrid at 4.8 kg empty (but includes VTOL motors and mounting). Pure CFRP frames like the SUX61 achieve 3.4 kg (Fact #11) at larger scale.
+
+### Conclusion
+**Primary CFRP (T700) with selective Kevlar reinforcement at impact zones** is optimal. The weight savings from CFRP directly translate to larger battery budget. T700 is preferred over T800 due to better impact tolerance and 44% lower cost at nearly identical density (Source #14). Kevlar layers at landing gear mounts and belly add crash protection without significant weight penalty (~100-200g).
+
+### Confidence
+✅ High — supported by multiple L1/L2 sources and confirmed by benchmark platforms.
+
+---
+
+## Dimension 2: Construction Method
+
+### Fact Confirmation
+Foam-core sandwich construction with CFRP skins achieves 30.5% wing weight reduction vs solid composite (Fact #4). Vacuum infusion produces 71% higher compressive strength than hand layup and 53% higher than vacuum bagging (Source #15). Prepreg achieves <0.5% void content but at higher cost.
+
+### Reference Comparison
+Industry benchmark platforms (Albatross, DeltaQuad) use composite sandwich construction. Academic research confirms sandwich panels with foam cores provide the highest stiffness-to-weight ratio for wing structures. Monocoque is preferred for fuselage sections where torsional loads dominate.
+
+### Conclusion
+**Sandwich construction (foam core + CFRP skin) for wings; monocoque for fuselage** is the optimal hybrid approach. Vacuum infusion is the recommended manufacturing process — best quality-to-cost ratio. Prepreg with autoclave cure would deliver superior results but requires expensive tooling; not justified for prototype phase.
+
+### Confidence
+✅ High — well-established in aerospace, confirmed by peer-reviewed research.
+
+---
+
+## Dimension 3: Foam Core Selection
+
+### Fact Confirmation
+Rohacell PMI has highest stiffness-to-weight and withstands autoclave temps (180°C+). XPS offers good closed-cell structure at low cost. PVC (Divinycell) is the industry standard middle ground. EPS is cheapest but has lowest strength and absorbs moisture (Source #16).
+
+### Reference Comparison
+For a UAV operating in -10°C to 45°C, thermal resistance beyond 80°C is sufficient (no autoclave required if using vacuum infusion). XPS and PVC both meet this requirement. Rohacell's premium is justified only for mass-produced military/aerospace where grams matter at scale.
+
+### Conclusion
+**PVC (Divinycell H-series) for wing cores** — best balance of stiffness, moisture resistance, and cost for prototype phase. XPS as budget alternative if cost optimization needed. Rohacell only justified if transitioning to production where marginal weight savings compound.
+
+### Confidence
+✅ High — PVC foam cores are industry standard for composite UAV wings.
+
+---
+
+## Dimension 4: Battery Technology
+
+### Fact Confirmation
+Semi-solid state batteries achieve 300-350 Wh/kg at cell level, ~310 Wh/kg at pack level (Fact #5, Source #20). Tattu 330Wh/kg 6S 33000mAh pack: 2324g weight, 732.6 Wh energy (Source #20). LiPo: 150-200 Wh/kg. Li-Ion: 200-250 Wh/kg. Semi-solid cycle life 800-1200 vs LiPo 200-500 (Fact #8, Source #17).
+
+### Endurance Estimate
+Reference: Albatross at 10 kg MTOW, ~3 kg LiPo (~180 Wh/kg = 540 Wh), gets 4 hours → cruise power ~135W.
+
+With semi-solid (same 3 kg at 310 Wh/kg = 930 Wh):
+Endurance = 930 Wh / 135 W ≈ 6.9 hours
+
+With optimized lighter airframe (saving 0.5 kg → extra battery weight):
+3.5 kg semi-solid at 310 Wh/kg = 1085 Wh
+Endurance = 1085 / 135 ≈ 8.0 hours
+
+The DeltaQuad Evo achieved 8h55m with solid-state batteries at 10 kg MTOW — validating the 6-8 hour range estimate for semi-solid + optimized airframe.
+
+### Conclusion
+**Semi-solid state batteries are the clear choice** for maximizing endurance. The Tattu 330 Wh/kg 6S or 12S packs are the most accessible commercial option. Expected endurance: 5-8 hours depending on airframe optimization and battery configuration. Cost premium (~2-3x LiPo per Wh) is offset by 4-6x cycle life and 30-50% more energy per kg.
+
+### Confidence
+✅ High — validated by DeltaQuad Evo real-world results and multiple manufacturer specifications.
+
+---
+
+## Dimension 5: Cost Analysis
+
+### Fact Confirmation
+Carbon fiber T700: ~$18/m². Custom composite airframe prototyping services available globally (Source #19). Tattu semi-solid batteries: ~$500-1500 per pack (estimated from capacity/chemistry). Total system costs for this class: $30-60k for first prototype.
+
+### Budget Breakdown (estimated for $100k)
+
+| Item | Estimated Cost |
+|------|---------------|
+| Airframe design + engineering | $10,000-15,000 |
+| Composite tooling + molds | $5,000-10,000 |
+| Materials (CF, foam, resin) | $3,000-5,000 |
+| Airframe manufacturing (outsourced) | $5,000-10,000 |
+| Motor, ESC, propeller | $1,000-2,000 |
+| Semi-solid batteries (2-3 packs) | $2,000-4,000 |
+| Avionics (Pixhawk 6x, GPS, telemetry) | Already owned |
+| Payload (camera, gimbal, Jetson) | Already owned |
+| Ground station + data link | $5,000-10,000 |
+| Integration + testing | $10,000-15,000 |
+| Contingency (~20%) | $10,000-15,000 |
+| **Total** | **$51,000-86,000** |
+
+### Conclusion
+$100k budget is sufficient with margin. CFRP airframe outsourcing is the most cost-effective path — avoids $50-100k investment in autoclave/clean-room equipment.
+
+### Confidence
+⚠️ Medium — cost estimates based on industry ranges; actual quotes needed from manufacturers.
+
+---
+
+## Dimension 6: Carbon Fiber Grade Selection
+
+### Fact Confirmation
+T700: tensile 4900 MPa, modulus 230 GPa, ~$18/m², better impact tolerance. T800: tensile 5880 MPa, modulus 294 GPa, ~$26/m², 44% cost premium, more brittle (Source #14). Density is nearly identical (~1.80 vs ~1.81 g/cm³).
+
+### Conclusion
+**T700 for all primary structure.** The 44% cost premium of T800 buys 15-20% more strength and 28% more stiffness, but T700 already exceeds structural requirements for this MTOW class. T800's brittleness is a liability for a UAV that may experience hard landings. T800 only justified for specific high-load areas (wing root spar caps) if FEA shows need.
+
+### Confidence
+✅ High — standard industry recommendation for UAV class.
@@ -0,0 +1,43 @@
+# Validation Log
+
+## Validation Scenario
+A fixed-wing reconnaissance UAV with 1.47 kg payload, targeting maximum endurance within 10 kg MTOW and $100k budget.
+
+Design: CFRP (T700) sandwich construction with PVC foam cores, powered by semi-solid state batteries (Tattu 330 Wh/kg).
+
+## Expected Based on Conclusions
+
+**Airframe weight**: 3.0-3.5 kg bare (based on Albatross benchmark at 3.35 kg with fiberglass+CF; pure CFRP should be lighter)
+
+**Battery allocation**: 3.0-4.0 kg (target 3.5 kg semi-solid at ~310 Wh/kg pack = ~1085 Wh)
+
+**Total weight**: 3.2 (airframe+avionics+motor) + 1.47 (payload) + 3.5 (battery) = 8.17 kg — under 10 kg MTOW with margin
+
+**Cruise power**: ~120-140 W (based on Albatross reference at ~135W for similar MTOW/speed)
+
+**Endurance**: 1085 Wh / 135 W ≈ 8.0 hours
+
+**Cost**: $50-85k total (within $100k budget)
+
+## Actual Validation Results
+Cross-checked against DeltaQuad Evo: 4.8 kg empty, 10 kg MTOW, 22 Ah × 22.2V × 2 = ~978 Wh, achieved 4h32m (standard) and 8h55m (solid-state record). Our lighter payload (1.47 kg vs 3 kg) and similar battery energy put us in the 6-8 hour range — consistent with DeltaQuad data.
+
+Albatross validation: 3.35 kg airframe, 10 kg MTOW, 4 hours with LiPo. Semi-solid upgrade alone (same airframe) would yield ~6.9 hours. With optimized CFRP airframe saving ~0.5 kg → additional battery weight → ~8 hours.
+
+## Counterexamples
+1. **Wind/turbulence**: Real-world endurance is typically 70-80% of theoretical due to wind, maneuvers, and non-optimal cruise segments. Realistic expectation: 5-6 hours practical endurance.
+2. **Battery degradation**: Semi-solid batteries lose capacity over cycles; after 500 cycles at 90% retention, endurance drops to ~5.4 hours.
+3. **Payload power draw**: Jetson Orin Nano Super draws ~15-25W, camera/gimbal ~10-15W. Total payload power: ~25-40W. This must be added to cruise power → total system power ~160-175W, reducing endurance to ~6.2-6.8 hours theoretical, ~5-5.5 hours practical.
+
+## Review Checklist
+- [x] Draft conclusions consistent with fact cards
+- [x] No important dimensions missed
+- [x] No over-extrapolation
+- [x] Conclusions actionable/verifiable
+- [x] Payload power consumption accounted for (see counterexample #3)
+
+## Conclusions Requiring Revision
+Endurance estimate revised downward from 8 hours theoretical to **5-6 hours practical** after accounting for:
+- Payload power draw (~30W)
+- Real-world flight efficiency (75%)
+- Battery reserve requirements (typically 20% reserve)
@@ -0,0 +1,177 @@
+# Solution Draft
+
+## Product Solution Description
+
+A custom-built electric fixed-wing reconnaissance UAV optimized for maximum flight endurance. The airframe uses **T700 carbon fiber composite sandwich construction** (CFRP skins over PVC foam cores for wings, CFRP monocoque for fuselage) with selective Kevlar reinforcement at impact zones. Powered by **semi-solid state batteries** (330 Wh/kg class), the platform carries a 1.47 kg reconnaissance payload (ADTI 20L V1 + Viewpro A40 Pro gimbal + Jetson Orin Nano Super + Pixhawk 6x).
+
+**Target performance**: 5-6 hours practical flight endurance, 8-10 kg MTOW, 2.5-3.5m wingspan.
+
+```
+┌─────────────────────────────────────────────────────────┐
+│                    SYSTEM OVERVIEW                       │
+│                                                         │
+│  CFRP Sandwich Wing (PVC foam core + T700 CF skin)      │
+│       ┌──────────────────────────────────┐              │
+│       │    High-aspect-ratio wing        │              │
+│       │    Wingspan: 3.0-3.2m            │              │
+│       └──────────┬───────────────────────┘              │
+│                  │                                       │
+│  ┌───────────────┴───────────────────┐                  │
+│  │   CFRP Monocoque Fuselage         │                  │
+│  │   ┌─────────┐  ┌──────────────┐   │                 │
+│  │   │ Battery │  │  Payload Bay │   │                  │
+│  │   │ Bay     │  │  (1.47 kg)   │   │                 │
+│  │   └─────────┘  └──────────────┘   │                  │
+│  └───────────────┬───────────────────┘                  │
+│                  │                                       │
+│          ┌───────┴───────┐                              │
+│          │  Motor + Prop  │                              │
+│          │  (pusher)      │                              │
+│          └───────────────┘                              │
+│                                                         │
+│  Power: Semi-solid state battery (Tattu 330Wh/kg)       │
+│  Avionics: Pixhawk 6x + GPS                            │
+│  Compute: Jetson Orin Nano Super                        │
+└─────────────────────────────────────────────────────────┘
+```
+
+## Existing/Competitor Solutions Analysis
+
+| Platform | MTOW | Endurance | Payload | Airframe Material | Battery | Price |
+|----------|------|-----------|---------|-------------------|---------|-------|
+| Applied Aeronautics Albatross | 10 kg | 4 hours | 4.5 kg | Fiberglass + Carbon fiber | LiPo | ~$8,000 (RTF) |
+| DeltaQuad Evo | 10 kg | 4h32m (std) / 8h55m (record) | 1-3 kg | Fiberglass + Carbon + Kevlar | Semi-solid / Solid-state Li | ~$25,000+ |
+| Penguin BE | <25 kg class | 110 min | 2.8 kg | Composite | Li-Ion | ~$30,000+ |
+| SUX61 | ~11 kg | 91 min | 8 kg | Carbon fiber monocoque | LiPo | ~$5,000 (frame) |
+
+**Key takeaway**: DeltaQuad Evo demonstrates that semi-solid/solid-state batteries combined with composite airframe can achieve 8+ hours in this MTOW class. Our design targets a similar approach with a lighter payload (1.47 vs 3 kg), leaving more weight budget for batteries.
+
+## Architecture
+
+### Component: Frame Material
+
+| Solution | Tools | Advantages | Limitations | Requirements | Security | Cost | Fit |
+|----------|-------|-----------|-------------|-------------|----------|------|-----|
+| **T700 CFRP (recommended)** | T700 unidirectional + woven prepreg or dry fabric | 40-50% lighter than Al, specific stiffness 113, excellent fatigue life, corrosion-proof | Brittle under impact, requires specialized manufacturing, difficult field repair | Vacuum infusion or prepreg + oven cure, outsourced manufacturing | N/A | ~$18/m² material; $15-25k total airframe manufacturing | ✅ Best for endurance |
+| Fiberglass (E-glass) | E-glass woven fabric + epoxy | Cheap (~$5/m²), easy to work, good impact tolerance, simple field repair | 40% heavier than CFRP for same stiffness, limits endurance | Basic workshop or outsource | N/A | ~$5/m²; $5-10k total | ⚠️ Weight penalty reduces endurance by ~1-2 hours |
+| Carbon-Kevlar Hybrid | Hybrid woven fabric | Best crash survivability, 25-40% lighter than Al | Kevlar hard to machine, UV sensitive, expensive (~$30/m²) | Specialized cutting tools, UV-protective coating | N/A | ~$30/m²; $20-30k total | ⚠️ Overkill for cost; Kevlar benefits limited to impact zones |
+| Aluminum 6061-T6 | CNC machining | Cheapest, easiest to manufacture, excellent repairability | Heaviest option (2.7 g/cm³), poor fatigue, reduces endurance 2-3 hours | CNC shop | N/A | ~$3-5k total | ❌ Weight kills endurance |
+
+**Recommendation**: T700 CFRP as primary structure with Kevlar patches at landing gear attach points and belly panel for crash protection (~100-200g weight addition).
+
+### Component: Construction Method
+
+| Solution | Tools | Advantages | Limitations | Requirements | Security | Cost | Fit |
+|----------|-------|-----------|-------------|-------------|----------|------|-----|
+| **Sandwich (foam core + CFRP skin) — recommended for wings** | PVC foam (Divinycell H60-H80), T700 fabric, vacuum infusion setup | Highest stiffness/weight ratio, 30% lighter than solid composite, excellent for wings | Requires quality core material, careful bonding | Vacuum pump, bagging film, infusion consumables | N/A | Core: ~$500-1000; total wing set: $5-8k | ✅ Best for wing endurance |
+| Monocoque (solid CFRP shell) — recommended for fuselage | CFRP prepreg or wet layup over male mold | Good torsional rigidity, smooth aerodynamic surface, compact | Heavier than sandwich for same stiffness, needs precise molds | Female or male molds, oven cure | N/A | Molds: $3-5k; layup: $2-3k | ✅ Best for fuselage |
+| Spar + Rib + Skin (traditional) | CNC-cut ribs, CF tube spars, film/fabric skin | Easy to prototype and modify, lightweight if well-designed | More labor-intensive, aerodynamic surface quality depends on skin | CNC router for ribs, CF tubes | N/A | $2-4k materials | ⚠️ Good for prototyping, inferior surface finish |
+
+**Recommendation**: Sandwich wings + monocoque fuselage. Outsource manufacturing to a composite prototyping service (e.g., Scabro Innovations, Refitech, or similar).
+
+### Component: Foam Core (for wing sandwich)
+
+| Solution | Tools | Advantages | Limitations | Requirements | Security | Cost | Fit |
+|----------|-------|-----------|-------------|-------------|----------|------|-----|
+| **PVC — Divinycell H60/H80 (recommended)** | Standard composite tools | Industry standard, good stiffness/weight, closed-cell moisture immune, handles 80°C cure | Not suitable for autoclave temps >100°C | Compatible with vacuum infusion and oven cure | N/A | ~$50-80/m² | ✅ Best value for prototype |
+| Rohacell PMI | Standard composite tools | Highest stiffness/weight, handles autoclave temps (180°C+) | Very expensive, overkill for prototype | Same as PVC | N/A | ~$150-300/m² | ⚠️ Only for production optimization |
+| XPS (extruded polystyrene) | Hot wire cutting | Cheapest, easy to shape, closed-cell | Lower compressive strength, limited to 75°C cure | Hot wire cutter | N/A | ~$10-20/m² | ⚠️ Budget option, acceptable for first prototype |
+| EPS (expanded polystyrene) | Hot wire cutting | Cheapest available | Lowest strength, absorbs moisture, open-cell-like bead structure | Hot wire cutter | N/A | ~$5-10/m² | ❌ Not recommended for flight-critical parts |
+
+### Component: Battery Technology
+
+| Solution | Tools | Advantages | Limitations | Requirements | Security | Cost | Fit |
+|----------|-------|-----------|-------------|-------------|----------|------|-----|
+| **Semi-solid state — Tattu 330Wh/kg (recommended)** | Compatible charger (6S/12S balance) | 310 Wh/kg pack level, 800-1200 cycles, -20 to 60°C, 10C peak | Higher cost per Wh (~$0.50-0.80), limited supplier options | Standard balance charger, battery management | Fire safety: low thermal runaway risk | ~$800-1500/pack (est.) | ✅ Best for max endurance |
+| Semi-solid state — Grepow 300Wh/kg | Compatible charger | 300 Wh/kg, 1200+ cycles, 2C charge, multiple configs | Slightly lower energy density than Tattu 330 | Standard balance charger | Fire safety: low risk | ~$700-1200/pack (est.) | ✅ Good alternative |
+| Li-Ion 21700 Pack (custom) | Spot welder, BMS, pack assembly | 200-250 Wh/kg, 500-800 cycles, widely available, cheap cells | Lower energy density, requires custom pack building, 3-5C max discharge | BMS, spot welder, cell matching | Medium: requires proper BMS | ~$0.20-0.35/Wh | ⚠️ 20-30% less endurance than semi-solid |
+| LiPo (traditional) | Standard RC charger | Cheapest, highest discharge rates (25-50C), widely available | 150-200 Wh/kg, 200-500 cycles, thermal sensitivity | Standard RC charger | Higher thermal runaway risk | ~$0.15-0.25/Wh | ❌ 40-50% less endurance than semi-solid |
+
+**Recommended configuration**: Tattu 330Wh/kg 6S 33000mAh × 1-2 packs (series or parallel depending on motor voltage requirements).
+- 1 pack: 2324g, 732.6 Wh → estimated 4-5 hours practical endurance
+- 2 packs (parallel): 4648g, 1465 Wh → estimated 6-7 hours practical (but may exceed MTOW)
+
+Optimal: single large 12S pack or purpose-selected configuration to stay within MTOW.
+
+### Component: Carbon Fiber Grade
+
+| Solution | Tools | Advantages | Limitations | Requirements | Security | Cost | Fit |
+|----------|-------|-----------|-------------|-------------|----------|------|-----|
+| **T700 (recommended)** | Standard composite tools | 4900 MPa tensile, 230 GPa modulus, good impact tolerance, industry standard for UAVs | Lower modulus than T800 | Standard resin systems | N/A | ~$18/m² | ✅ Best value |
+| T800 | Standard composite tools | 5880 MPa tensile, 294 GPa modulus, 28% stiffer | 44% more expensive, more brittle, marginal weight gain at this scale | Same resin systems | N/A | ~$26/m² | ⚠️ Only for specific high-load elements |
+| T300 | Standard composite tools | Cheapest, widely available | Significantly lower strength than T700 | Same resin systems | N/A | ~$12/m² | ❌ Insufficient for primary structure |
+
+## Weight Budget Estimate
+
+| Component | Weight (kg) |
+|-----------|-------------|
+| Bare airframe (CFRP sandwich wing + monocoque fuselage) | 2.8-3.2 |
+| Motor + ESC + propeller | 0.4-0.6 |
+| Wiring, connectors, misc hardware | 0.3-0.5 |
+| Payload (camera + gimbal + Jetson + Pixhawk + GPS) | 1.47 |
+| Battery (semi-solid, target) | 3.0-3.5 |
+| **Total estimated** | **8.0-9.3** |
+| MTOW limit | 10.0 |
+| **Margin** | **0.7-2.0** |
+
+## Endurance Estimate
+
+**Assumptions**:
+- MTOW: 9.0 kg (mid-range estimate)
+- Cruise speed: 17 m/s
+- L/D ratio: ~15 (high-aspect-ratio wing)
+- Propulsive efficiency: 0.85
+- Battery: 3.2 kg semi-solid at 310 Wh/kg = 992 Wh
+- Payload power: ~30W (Jetson 15-25W + camera/gimbal 10-15W)
+- Cruise power: ~130W (aerodynamic) + 30W (payload) = ~160W total
+- Battery reserve: 20%
+- Usable energy: 992 × 0.80 = 794 Wh
+
+**Theoretical endurance**: 992 / 160 = 6.2 hours
+**Practical endurance (with reserve + real-world losses)**: 794 / 160 ≈ **5.0 hours**
+
+**Range at cruise**: 5.0h × 17 m/s × 3.6 = **306 km**
+
+This is conservative. Optimization of airfoil, wing loading, and propulsion system could push practical endurance to 5.5-6.0 hours.
+
+## Testing Strategy
+
+### Integration / Functional Tests
+- Static load test: wing spar to 3× max flight load (verify no failure at 3g)
+- Ground vibration test: verify no flutter modes within flight envelope
+- Range/endurance test: fly at cruise speed until 20% battery reserve, measure actual endurance vs predicted
+- Payload integration test: verify all electronics (Jetson, Pixhawk, camera, gimbal) function correctly with airframe vibration
+- CG range test: verify stable flight across full CG envelope
+
+### Non-Functional Tests
+- Temperature endurance: ground soak at -10°C and +45°C, verify battery and avionics function
+- Wind resistance: fly in 10-12 m/s sustained wind, verify controllability and endurance impact
+- Hard landing test: drop from 1m at 2 m/s descent rate onto belly, verify structural integrity (Kevlar reinforcement zones)
+- Battery cycle test: charge/discharge 50 cycles, verify capacity retention ≥95%
+- EMI test: verify Jetson/camera does not interfere with GPS/telemetry
+
+## References
+
+1. UAVMODEL — Carbon Fiber Fixed Wing Drones: https://www.uavmodel.com/blogs/news/skyeye-sr260-fixed-wing-drone-2600mm-long-endurance-mapping-amp-inspection
+2. SUX61 UAV Frame: https://aerojetparts.com/product/sux61-uav-frame-carbon-fiber-8kg-payload-91min-endurance/
+3. FAI — Vanilla UAV Flight Duration Record: https://www.fai.org/vanilla-uav-flight-duration-record
+4. Springer — EPS-Fiber-Reinforced Composite Wing Analysis (2024): https://link.springer.com/10.1007/s11029-024-10185-3
+5. Grepow Semi-Solid Battery: https://www.grepow.com/semi-solid-state-battery/300wh-kg-series-high-energy-density-battery-pack.html
+6. Tattu Semi-Solid Battery: https://tattuworld.com/semi-solid-state-battery/
+7. Herewin Semi-Solid Guide (2026): https://www.herewinpower.com/blog/solid-state-drone-batteries-ultimate-guide/
+8. Applied Aeronautics Albatross: https://www.appliedaeronautics.com/albatross-uav
+9. KingRaysCarbon — CF vs Al: https://kingrayscarbon.com/carbon-fiber-vs-aluminum-for-drone-frames-which-performs-better/
+10. Dronecarbon — Kevlar vs CF: https://www.dronecarbon.com/kevlar-vs-carbon-fiber_a9075.html
+11. Herewin — LFP vs LiPo vs Semi-Solid (2026): https://www.herewinpower.com/blog/lfp-vs-lipo-vs-semi-solid-industrial-drone-batteries-2026-roi-safety-and-performance/
+12. DeltaQuad Evo Specs: https://docs.deltaquad.com/gov/vehicle-specifications
+13. DeltaQuad Evo 8h55m Record: https://uasweekly.com/2025/06/27/deltaquad-evo-sets-record-with-8-hour-flight-endurance-for-electric-vtol-uas-milestone/
+14. T700 vs T800 Guide: https://www.carbonfibermaterial.com/t700-vs-t800-carbon-fiber-a-practical-guide-for-material-selection/
+15. CFRP Manufacturing Comparison (Indonesian J. Aerospace): https://ejournal.brin.go.id/ijoa/article/view/286
+16. Rohacell vs Foam Cores — Chem-Craft: https://chem-craft.com/blog/comparative-analysis-rohacell-vs-traditional-materials-in-composite-engineering/
+17. Carbon-Kevlar Hybrid: https://ictmaterial.com/what-is-carbon-kevlar-hybrid-fabric-properties-and-use-cases/
+18. Scabro Innovations — UAV Prototyping: https://scabroinnovations.com/diensten/composite-airframe-prototyping/
+19. Tattu 330Wh/kg 6S Specs: https://www.tattuworld.com/semi-solid-state-battery/semi-solid-330wh-kg-33000mah-22-2v-10c-6s-battery.html
+20. ASTM F3563-22: https://www.astm.org/f3563-22.html
+
+## Related Artifacts
+- AC Assessment: `_standalone/UAV_frame_material/00_research/UAV_frame_material/00_ac_assessment.md`
@@ -0,0 +1,428 @@
+# Solution Draft (Rev 02)
+
+## Revised Constraints (vs Draft 01)
+
+| Constraint | Draft 01 | Draft 02 |
+|-----------|----------|----------|
+| Cost per unit | $100k prototype | < $7k, target < $5k |
+| Material | CFRP (T700) | S2 fiberglass (radio transparent) |
+| Radio transparency | Not considered | Required — full RF transparency for GPS, telemetry, data links |
+| Flight time | 5-6 hours target | Same if possible, can be less |
+| Transport | Not specified | Disassembled fits in car trunk; 2 planes per pickup truck |
+
+## Product Solution Description
+
+A modular, radio-transparent electric fixed-wing reconnaissance UAV built with **S2 fiberglass/foam-core sandwich construction** with internal **carbon fiber spar reinforcement**. Designed for field deployment — disassembles into 3 sections (2 wing panels + fuselage) that fit in a car trunk, with 2 complete aircraft fitting in a standard pickup truck bed. Powered by semi-solid state batteries for maximum endurance.
+
+**Target performance**: 3.5-5 hours practical flight endurance, 9-10 kg MTOW, ~3m wingspan, < $5k BOM per unit.
+
+```
+┌──────────────────────────────────────────────────────────────┐
+│                  MODULAR AIRFRAME LAYOUT                      │
+│                                                              │
+│   LEFT WING PANEL        FUSELAGE         RIGHT WING PANEL   │
+│   (~1.5m span)          (~1.0-1.1m)       (~1.5m span)      │
+│  ┌──────────────┐   ┌──────────────┐   ┌──────────────┐     │
+│  │ S2 FG skin   │   │ S2 FG skin   │   │ S2 FG skin   │     │
+│  │ PVC foam core│◄─►│ Battery bay  │◄─►│ PVC foam core│     │
+│  │ CF spar cap  │   │ Payload bay  │   │ CF spar cap  │     │
+│  │ (internal)   │   │ Motor+ESC    │   │ (internal)   │     │
+│  └──────────────┘   └──────────────┘   └──────────────┘     │
+│                                                              │
+│  Wing-fuselage joint: aluminum spar joiner + 2 pin locks     │
+│  Assembly time target: < 10 minutes                          │
+│  Material: S2 fiberglass = RF transparent (GPS/telemetry OK) │
+│  Internal CF spar: minimal RF impact (narrow linear element)  │
+└──────────────────────────────────────────────────────────────┘
+
+TRANSPORT CONFIGURATION (standard pickup truck, 6.5ft bed):
+┌───────────────────────────────────────────┐
+│  Truck bed: 198cm × 130cm (wheel wells)   │
+│  ┌──────────────────┐ ┌────────────────┐  │
+│  │ Plane 1 wings    │ │ Plane 2 wings  │  │
+│  │ (2 × 150cm long) │ │ (2 × 150cm)    │  │
+│  │ stacked ~20cm    │ │ stacked ~20cm  │  │
+│  ├──────────────────┤ ├────────────────┤  │
+│  │ Plane 1 fuselage │ │ Plane 2 fuse.  │  │
+│  │ (~110cm)         │ │ (~110cm)       │  │
+│  └──────────────────┘ └────────────────┘  │
+│  Total width: ~60cm × 2 = 120cm < 130cm ✓│
+│  Total length: 150cm < 198cm ✓            │
+└───────────────────────────────────────────┘
+```
+
+## Existing/Competitor Solutions Analysis
+
+| Platform | MTOW | Endurance | Payload | Material | RF Transparent | Modular | Price |
+|----------|------|-----------|---------|----------|---------------|---------|-------|
+| Albatross (kit) | 10 kg | 4 hours | 4.5 kg | Fiberglass + CF | Partial | No (removable wings) | $1,500 kit |
+| Albatross (RTF) | 10 kg | 4 hours | 4.5 kg | Fiberglass + CF | Partial | No | $4,800 |
+| DeltaQuad Evo | 10 kg | 4.5h / 8.9h record | 1-3 kg | FG + CF + Kevlar | Partial | Wing removable | $25,000+ |
+| Skywalker X8 | ~4 kg | 45-60 min | 1-2 kg | EPO foam | Yes | No | $489-598 |
+| Mugin 2600 | 15 kg | 1.5-5h | 4 kg | Carbon fiber | No | Wing sections | $2,299+ |
+
+**Key insight**: The Albatross kit at $1,500 proves that a 3m wingspan composite airframe is achievable at very low cost. Our target of < $5k per complete unit (with batteries) is realistic. No competitor offers the combination of radio transparency + modular transport + semi-solid batteries.
+
+## Architecture
+
+### Component: Frame Material
+
+| Solution | Advantages | Limitations | Cost (per unit) | Fit |
+|----------|-----------|-------------|----------------|-----|
+| **S2 fiberglass skin + PVC foam core + internal CF spar (recommended)** | RF transparent skin, strong internal structure, good impact tolerance, easy to repair, $8-19/m² fabric | ~30-40% heavier than pure CFRP for equivalent stiffness | Fabric: $200-400; foam: $100-200; CF spar material: $50-100; resin: $80-150; total materials: $430-850 | ✅ Best balance of RF transparency, cost, repairability |
+| E-glass fiberglass (instead of S2) | Cheapest glass option (~$3-5/m²), RF transparent, easy to work | 40% weaker than S2, requires thicker layup → heavier | Materials: $200-500 | ⚠️ Acceptable budget option, slightly heavier |
+| Pure S2 fiberglass (no CF spar) | Maximum RF transparency, simplest construction | Insufficient wing stiffness at low weight, flutter risk | Materials: $300-600 | ❌ Stiffness deficit at acceptable weight |
+| Pure CFRP (draft 01 approach) | Lightest, stiffest | Blocks RF — GPS/telemetry degraded, expensive | Materials: $800-1500 | ❌ Fails radio transparency requirement |
+
+**Recommendation**: S2 fiberglass skin over PVC foam core with unidirectional carbon fiber spar caps (top and bottom of main spar, internal). The CF spar is a narrow linear element (~20-30mm wide per cap) inside the wing — negligible RF blockage. All external surfaces are S2 FG = fully radio transparent.
+
+### Component: Construction Method
+
+| Solution | Advantages | Limitations | Cost | Fit |
+|----------|-----------|-------------|------|-----|
+| **Vacuum-bagged foam sandwich (recommended)** | Good quality (53% stronger than hand layup), low tooling cost, reproducible | Requires vacuum pump + consumables | Equipment: $500 one-time; consumables: $50-100/unit | ✅ Best for low-cost production |
+| Hand layup over foam core | Cheapest, simplest, no equipment needed | Lower quality (more voids), less consistent | Minimal equipment | ⚠️ Acceptable for prototypes only |
+| Vacuum infusion | Best quality (71% stronger than hand layup) | More complex setup, higher consumable cost | Equipment: $1000+; consumables: $100-200/unit | ⚠️ Worth it at higher volume (>20 units) |
+| Outsourced prepreg manufacturing | Highest quality | Expensive per unit at low volume | $2000-5000/airframe | ❌ Exceeds per-unit budget |
+
+### Component: Foam Core
+
+| Solution | Advantages | Limitations | Cost/m² | Fit |
+|----------|-----------|-------------|---------|-----|
+| **PVC Divinycell H60 (recommended)** | Good stiffness/weight, closed-cell, 80°C tolerant, industry standard | More expensive than XPS | $50-80/m² | ✅ Best value for production |
+| XPS (extruded polystyrene) | Cheapest closed-cell, easy to shape with hot wire | Lower compressive strength, 75°C limit | $10-20/m² | ✅ Good budget alternative |
+| EPS (expanded polystyrene) | Very cheap | Absorbs moisture, lowest strength | $5-10/m² | ⚠️ Only for non-critical areas |
+
+### Component: Wing-Fuselage Joint (Modular Assembly)
+
+| Solution | Advantages | Limitations | Cost | Fit |
+|----------|-----------|-------------|------|-----|
+| **Aluminum spar joiner + pin locks (recommended)** | Quick assembly (<5 min), proven in RC/UAV, high strength, replaceable | Adds ~100-150g per joint (200-300g total) | $30-60 machined aluminum parts | ✅ Simple, reliable, fast |
+| 3D-printed spar connector with hinge | Very fast assembly (<2 min), lightweight | Lower strength, fatigue concerns, requires testing | $10-20 per set | ⚠️ Good for prototype, risky for production |
+| Bolted flange joint | Very strong, proven in full-scale aviation | Heavier (~200g per joint), slower assembly (10+ min) | $20-40 | ⚠️ Over-engineered for this scale |
+
+**Design**: Wing spar is a carbon fiber tube or C-channel running the full wing half-span. At the root, it slides into an aluminum joiner tube embedded in the fuselage. Secured with 2 quick-release pins (top and bottom). Electrical connections (servo leads) via a quick-disconnect plug at each wing root.
+
+### Component: Battery Technology
+
+| Solution | Energy density | Endurance impact | Cycle life | Cost/pack | Fit |
+|----------|---------------|-----------------|------------|-----------|-----|
+| **Semi-solid Tattu 330Wh/kg 6S (recommended)** | 315 Wh/kg pack | Baseline (best) | 800-1200 | ~$800-1200 est. | ✅ Best endurance per $ |
+| Semi-solid Grepow 300Wh/kg 6S | 280-300 Wh/kg pack | -5 to -10% | 1200+ | ~$700-1000 est. | ✅ Good alternative |
+| Li-Ion 21700 custom pack (6S) | 200-220 Wh/kg pack | -25 to -35% | 500-800 | ~$200-400 | ⚠️ Budget option, significant endurance loss |
+| LiPo 6S (standard RC) | 150-180 Wh/kg pack | -40 to -50% | 200-500 | ~$100-200 | ❌ Too much endurance loss |
+
+## Weight Budget (S2 Fiberglass Build)
+
+| Component | Weight (kg) | Notes |
+|-----------|-------------|-------|
+| Bare airframe (S2 FG sandwich + CF spar) | 3.8-4.5 | ~30% heavier than pure CFRP; Albatross FG+CF is 3.35 kg |
+| Wing joints (aluminum) | 0.2-0.3 | Spar joiners + pins + quick-disconnect plugs |
+| Motor + ESC + propeller | 0.4-0.6 | |
+| Wiring, connectors, misc | 0.3-0.4 | |
+| **Platform subtotal** | **4.7-5.8** | |
+| Payload (camera + gimbal + Jetson + Pixhawk + GPS) | 1.47 | Fixed |
+| Battery (semi-solid) | 2.7-3.8 | Remainder to MTOW |
+| **Total (target MTOW 10 kg)** | **~10.0** | |
+
+Conservative estimate: platform 5.3 kg + payload 1.47 kg + battery 3.2 kg = 9.97 kg.
+
+## Endurance Estimate (S2 Fiberglass)
+
+**Assumptions**:
+- MTOW: 10 kg
+- Platform weight: 5.3 kg (S2 FG airframe + motor + wiring + joints)
+- Payload: 1.47 kg
+- Battery: 3.23 kg semi-solid at 310 Wh/kg = 1001 Wh
+- Cruise power: ~140W (slightly higher than CFRP due to heavier aircraft → higher induced drag)
+- Payload power: ~30W (Jetson + camera + gimbal)
+- Total system power: ~170W
+- Battery reserve: 20%
+- Usable energy: 1001 × 0.80 = 801 Wh
+- Real-world efficiency factor: 0.75
+
+**Theoretical endurance**: 1001 / 170 = 5.9 hours
+**Practical endurance (with reserve)**: 801 / 170 ≈ **4.7 hours**
+**Practical endurance (with reserve + real-world losses)**: 801 × 0.75 / 170 ≈ **3.5 hours**
+
+**Comparison to Draft 01 (CFRP)**:
+- Draft 01: 5.0 hours practical → Draft 02: 3.5-4.7 hours practical
+- Endurance reduction: ~15-30% depending on conditions
+- Still competitive with Albatross (4h with LiPo) when using semi-solid batteries
+
+**With budget Li-Ion pack instead** (to stay under $5k):
+- 3.23 kg Li-Ion at 210 Wh/kg = 678 Wh → usable 542 Wh
+- Practical: 542 / 170 ≈ **3.2 hours** (reserve only) / **2.4 hours** (worst case)
+
+## BOM Cost Estimate (Per Unit)
+
+| Component | Low Est. | High Est. | Notes |
+|-----------|----------|-----------|-------|
+| S2 fiberglass fabric | $150 | $300 | ~8-10 m² at $15-30/m² |
+| PVC foam core (Divinycell H60) | $100 | $200 | Wing + fuselage panels |
+| Epoxy resin + hardener | $80 | $150 | ~2-3 kg resin |
+| CF spar material (tube + UD tape) | $50 | $100 | Spar caps + tubes |
+| Aluminum spar joiners (machined) | $30 | $60 | 2 joiner sets, batch machined |
+| Vacuum bagging consumables | $30 | $60 | Bag, breather, peel ply, tape |
+| Motor (brushless, ~500W) | $80 | $150 | |
+| ESC (40-60A) | $40 | $80 | |
+| Propeller (folding) | $15 | $30 | |
+| Servos (4× ailerons + elevator + rudder) | $60 | $120 | |
+| Wiring, connectors, hardware | $50 | $100 | |
+| Semi-solid battery (Tattu 330Wh/kg 6S 33Ah) | $800 | $1,200 | Single pack |
+| RC receiver | $30 | $80 | |
+| Telemetry radio | $100 | $300 | |
+| Transport case / padded bag | $50 | $150 | |
+| **Subtotal (airframe + propulsion + battery)** | **$1,665** | **$3,080** | |
+| Pixhawk 6x + GPS | $300 | $500 | If not already owned |
+| **Total BOM (without mission payload)** | **$1,965** | **$3,580** | |
+| **Total BOM (with Pixhawk, without mission payload)** | **$2,265** | **$4,080** | |
+
+Manufacturing labor (per unit, assuming in-house build with molds amortized):
+- First unit (mold making): +$2,000-3,000 tooling
+- Subsequent units: ~$500-1,000 labor per airframe (8-16 hours assembly)
+
+**Per-unit cost at batch of 5+**: **$2,800-4,500** (without mission payload) ✅ Under $5k target
+**Per-unit cost at batch of 1 (first prototype)**: **$5,000-7,000** (includes tooling) ✅ Under $7k target
+
+## Modular Transport Specifications
+
+| Dimension | Value |
+|-----------|-------|
+| Wing panel length | ~1.50 m (half-span) |
+| Wing panel chord | ~0.25-0.30 m |
+| Wing panel thickness | ~0.04-0.05 m |
+| Fuselage length | ~1.00-1.10 m |
+| Fuselage width/height | ~0.15-0.20 m |
+| Assembly time | < 10 minutes (target) |
+| Disassembly time | < 5 minutes |
+
+**Car trunk fit**: 3 sections (2 wings + fuselage) fit in standard sedan trunk (~120×80×45 cm). Wings stack flat, fuselage alongside. ✅
+
+**Pickup truck (2 planes)**: Standard 6.5ft bed (198×130 cm between wheel wells). Each plane's longest component is 150 cm (< 198 cm bed length). Two planes side by side need ~120 cm width (< 130 cm between wheel wells). ✅
+
+## Trade-off Summary: S2 Fiberglass vs CFRP
+
+| Dimension | S2 Fiberglass (Draft 02) | CFRP (Draft 01) | Winner |
+|-----------|--------------------------|-----------------|--------|
+| RF transparency | ✅ Excellent — transparent to GPS, telemetry, data links | ❌ Blocks RF, requires external antennas | S2 FG |
+| Cost per unit | $2,800-4,500 | $30,000-60,000 (prototype) | S2 FG |
+| Endurance | 3.5-4.7 hours practical | 5.0 hours practical | CFRP (+15-30%) |
+| Airframe weight | 3.8-4.5 kg bare | 2.8-3.2 kg bare | CFRP (-25%) |
+| Impact resistance | Good (fiberglass is tough) | Poor (CFRP is brittle) | S2 FG |
+| Field repairability | Easy (fiberglass patches, epoxy) | Difficult (specialized repair) | S2 FG |
+| Manufacturing complexity | Low (basic vacuum bagging) | Medium-High (precise layup) | S2 FG |
+| Transport / modularity | Same | Same | Tie |
+
+**Conclusion**: S2 fiberglass is the clear choice given the revised constraints. The 15-30% endurance reduction vs CFRP is offset by radio transparency (critical for the mission), 10x lower cost, and significantly easier manufacturing and field repair.
+
+## Testing Strategy
+
+### Integration / Functional Tests
+- Static wing load test: 3× max flight load at spar joiner (verify no failure at 3g)
+- Wing joint cycling: 100× assembly/disassembly, verify no wear or looseness
+- RF transparency test: measure GPS signal strength through airframe skin vs free-air (target: < 3 dB attenuation)
+- Assembly time test: verify < 10 minutes from transport case to flight-ready
+- Range/endurance test: fly at cruise until 20% reserve, measure actual vs predicted
+- Payload integration test: all electronics function under vibration
+
+### Non-Functional Tests
+- Transport test: load 2 planes in pickup truck, drive 100 km on mixed roads, verify no damage
+- Hard landing test: belly landing at 2 m/s descent, verify structural integrity
+- Field repair test: simulate wing skin puncture, repair with FG patch + epoxy, verify airworthy in < 30 minutes
+- Temperature test: battery + avionics function at -10°C and +45°C
+- Battery cycle test: 50 charge/discharge cycles, verify ≥95% capacity retention
+
+## Production BOM: 5 UAVs From Scratch
+
+### A. One-Time Equipment & Tooling
+
+| Item | Qty | Unit Price | Total | Notes |
+|------|-----|-----------|-------|-------|
+| **Composite Workshop Equipment** | | | | |
+| Vacuum pump (6 CFM 2-stage) | 1 | $280 | $280 | VIOT or equivalent |
+| Vacuum bagging starter kit (gauges, tubing, valves, connectors) | 1 | $150 | $150 | |
+| Digital scale (0.1g precision, 5 kg capacity) | 1 | $50 | $50 | For resin mixing |
+| Mixing cups, squeegees, rollers, brushes set | 1 | $80 | $80 | |
+| Large work table (4×8 ft plywood + sawhorses) | 1 | $150 | $150 | |
+| Self-healing cutting mat (4×8 ft) | 1 | $80 | $80 | |
+| **Foam Cutting** | | | | |
+| CNC hot wire foam cutter (4-axis, DIY kit) | 1 | $350 | $350 | Vortex-RC or similar |
+| **Mold Making** | | | | |
+| MDF sheets for plugs (4×8 ft × ¾") | 4 | $45 | $180 | Wing + fuselage plugs |
+| Tooling epoxy + fiberglass for female molds | 1 | $600 | $600 | 2× wing mold halves + fuselage molds |
+| Mold release agent (PVA + wax) | 1 | $60 | $60 | |
+| Filler / fairing compound | 1 | $80 | $80 | For plug finishing |
+| Sandpaper assortment (80-600 grit) | 1 | $40 | $40 | |
+| **Metal Work** | | | | |
+| Aluminum spar joiner machining (batch of 12 sets) | 1 | $400 | $400 | CNC outsourced, 10 sets + 2 spare |
+| **PPE & Ventilation** | | | | |
+| Respirator (half-face, organic vapor + P100) | 2 | $40 | $80 | 1 per worker |
+| Nitrile gloves (box of 200) | 2 | $25 | $50 | |
+| Safety glasses | 3 | $10 | $30 | |
+| Portable fume extractor / fan | 1 | $120 | $120 | |
+| **Hand & Power Tools** | | | | |
+| Drill + mixing paddle | 1 | $80 | $80 | |
+| Jigsaw | 1 | $60 | $60 | |
+| Rotary tool (Dremel) | 1 | $50 | $50 | |
+| Heat gun | 1 | $35 | $35 | |
+| Scissors, utility knives, rulers, clamps | 1 | $80 | $80 | Assorted set |
+| **Charging & Testing** | | | | |
+| Battery charger (6S/12S balance, 1000W) | 1 | $200 | $200 | |
+| Multimeter | 1 | $30 | $30 | |
+| Servo tester | 1 | $15 | $15 | |
+| **Software & Design** | | | | |
+| CAD/CAM (FreeCAD / OpenVSP — free) | — | $0 | $0 | Open source |
+| Hot wire CNC software (included with cutter) | — | $0 | $0 | |
+| | | | | |
+| **EQUIPMENT & TOOLING TOTAL** | | | **$3,335** | |
+
+### B. Raw Materials (for 5 UAVs + 20% waste margin)
+
+Material quantities per UAV:
+- Wing skin area: ~1.6 m² planform × 2 (top+bottom) × 2 layers = ~6.4 m² S2 fabric
+- Fuselage skin: ~0.6 m² × 2 layers = ~1.2 m²
+- Tail surfaces: ~0.3 m² × 2 layers = ~0.6 m²
+- Total S2 fabric per UAV: ~8.2 m² → with waste: ~10 m²
+- Foam core per UAV: ~2.5 m² (wings + tail)
+- Resin per UAV: ~2.5 kg (fabric weight × 1:1 ratio + extra)
+
+| Item | Qty (5 UAVs + margin) | Unit Price | Total | Notes |
+|------|----------------------|-----------|-------|-------|
+| **Structural Materials** | | | | |
+| S2 fiberglass fabric 6oz (30" wide) | 70 yards (~64 m) | $12.50/yard | $875 | ~10 m² per UAV × 5 + waste |
+| PVC foam Divinycell H60 10mm (1.22×0.81m sheets) | 16 sheets | $40/sheet | $640 | ~2.5 m² per UAV × 5 + waste |
+| Laminating epoxy resin (West System 105 or equiv) | 4 gallons | $125/gal | $500 | ~2.5 kg resin per UAV |
+| Epoxy hardener | 2 gallons | $80/gal | $160 | |
+| Carbon fiber tube (spar, 20mm OD, 1.5m) | 12 | $25 each | $300 | 2 per UAV + spare |
+| Carbon fiber UD tape (spar caps, 25mm wide) | 30 m | $5/m | $150 | 5m per UAV + spare |
+| **Vacuum Bagging Consumables** | | | | |
+| Vacuum bag film (5m × 1.5m rolls) | 6 rolls | $20/roll | $120 | ~1 roll per UAV + spare |
+| Peel ply fabric | 20 yards | $5/yard | $100 | |
+| Breather cloth | 20 yards | $4/yard | $80 | |
+| Sealant tape | 6 rolls | $12/roll | $72 | |
+| **Hardware (per 5 UAVs)** | | | | |
+| Aluminum spar joiners | (included in tooling) | — | $0 | Batch machined above |
+| Quick-release pins (stainless) | 20 | $3 each | $60 | 4 per UAV |
+| Quick-disconnect electrical plugs | 10 | $8 each | $80 | 2 per UAV (wing roots) |
+| Misc hardware (bolts, nuts, hinges, control horns) | 5 sets | $30/set | $150 | |
+| | | | | |
+| **RAW MATERIALS TOTAL (5 UAVs)** | | | **$3,287** | |
+| **Per UAV materials** | | | **~$657** | |
+
+### C. Electronics & Propulsion (per UAV × 5)
+
+| Item | Qty/UAV | Unit Price | Per UAV | ×5 Total | Notes |
+|------|---------|-----------|---------|----------|-------|
+| Motor (brushless ~500W, e.g. Dualsky XM5050EA) | 1 | $90 | $90 | $450 | Fixed-wing optimized |
+| ESC (40-60A, BLHeli) | 1 | $50 | $50 | $250 | |
+| Folding propeller (13×8 or similar) | 2 | $15 | $30 | $150 | 1 spare per UAV |
+| Servos (digital metal gear, 15-20 kg·cm) | 5 | $25 | $125 | $625 | 2× aileron + elevator + rudder + flap/spare |
+| Pixhawk 6X Mini + GPS | 1 | $380 | $380 | $1,900 | |
+| RC receiver (long range, e.g. TBS Crossfire) | 1 | $60 | $60 | $300 | |
+| RFD900x telemetry pair (shared GCS unit) | 1 air + 0.2 GCS | $170 (air) | $170 | $850 + $350 GCS = $1,200 | 1 GCS module shared |
+| Power distribution board + BEC | 1 | $25 | $25 | $125 | |
+| Wiring, connectors (XT90, JST, servo ext.) | 1 set | $40 | $40 | $200 | |
+| Semi-solid battery (Tattu 330Wh/kg 6S 33Ah) | 1 | $732 | $732 | $3,660 | |
+| | | | | | |
+| **ELECTRONICS TOTAL (5 UAVs)** | | | | **$8,910** | |
+| **Per UAV electronics** | | | **~$1,702** | | Excl. shared GCS telemetry |
+
+### D. Consumables & Misc (for 5 UAVs)
+
+| Item | Total | Notes |
+|------|-------|-------|
+| Transport bags / padded cases (per UAV) | $300 | $60 × 5 (padded wing bags + fuselage bag) |
+| Battery charger cables + adapters | $50 | |
+| Field repair kit (S2 FG patches, epoxy sachets, sandpaper) | $150 | $30 × 5 |
+| Spare hardware kit (pins, bolts, servo horns) | $100 | |
+| Shipping / freight (materials + components) | $400 | Estimate |
+| **CONSUMABLES TOTAL** | **$1,000** | |
+
+### E. Labor
+
+| Role | People | Duration | Rate | Total | Notes |
+|------|--------|----------|------|-------|-------|
+| Mold making + setup (one-time) | 2 | 3 weeks | $30/hr | $7,200 | 2 people × 40h/wk × 3 wk |
+| Airframe layup + cure (per UAV) | 2 | 3 days | $30/hr | $2,880 | 2 people × 8h × 3 days × 5 UAVs |
+| Post-cure trim, finish, assembly | 1 | 2 days | $30/hr | $2,400 | 1 person × 8h × 2 days × 5 |
+| Electronics integration + wiring | 1 | 1.5 days | $35/hr | $2,100 | 1 person × 8h × 1.5 days × 5 |
+| QA, testing, calibration | 1 | 1 day | $35/hr | $1,400 | 1 person × 8h × 1 day × 5 |
+| **LABOR TOTAL** | | | | **$15,980** | |
+| **Per UAV labor** | | | | **~$2,516** | Including amortized mold making |
+
+### F. Production Summary — Total Investment for 5 UAVs
+
+| Category | Total | Per UAV |
+|----------|-------|---------|
+| A. Equipment & Tooling (one-time) | $3,335 | $667 |
+| B. Raw Materials | $3,287 | $657 |
+| C. Electronics & Propulsion | $8,910 | $1,782 |
+| D. Consumables & Misc | $1,000 | $200 |
+| E. Labor | $15,980 | $3,196 |
+| | | |
+| **GRAND TOTAL (5 UAVs)** | **$32,512** | |
+| **Per UAV (all-in, including labor)** | | **$6,502** |
+| **Per UAV (materials + electronics only, no labor)** | | **$3,306** |
+
+### G. Cost Optimization Options
+
+| Optimization | Savings/UAV | Impact |
+|-------------|-------------|--------|
+| Use XPS foam instead of Divinycell H60 | -$90 | Slightly lower stiffness, acceptable for prototype |
+| Use E-glass instead of S2 glass | -$100 | ~40% weaker, needs thicker layup → ~200g heavier |
+| Use Li-Ion 21700 pack instead of Tattu semi-solid | -$400 | Endurance drops from 3.5-4.7h to 2.4-3.2h |
+| Self-machine spar joiners (manual lathe) | -$50 | Requires metalworking skill |
+| Use cheaper servos ($15 each) | -$50 | Lower torque, shorter lifespan |
+| **Aggressive budget build** | **-$690** | **$2,616/UAV materials only** |
+
+### H. Minimum Viable Team
+
+| Role | Count | Skills Required | Commitment |
+|------|-------|----------------|------------|
+| Composite fabricator | 1-2 | Fiberglass layup, vacuum bagging, mold making | Full-time during build (8 weeks) |
+| Electronics/avionics tech | 1 | Soldering, Pixhawk configuration, wiring | Part-time (can overlap with fabricator) |
+| **Minimum: 2 people for 8 weeks** | | | |
+
+**Timeline for 5 UAVs**:
+- Week 1-3: Mold making (CNC foam plugs → fiberglass female molds)
+- Week 4-5: First 2 airframes layup + cure + trim
+- Week 5-6: Next 3 airframes layup + cure + trim
+- Week 6-7: Electronics integration all 5 units
+- Week 7-8: Testing, calibration, flight testing
+- **Total: ~8 weeks with 2 people**
+
+### I. Minimal Absolute Cost (No Labor Accounted)
+
+If labor is free (owner-operators building their own):
+
+| Category | Total | Per UAV |
+|----------|-------|---------|
+| Equipment & Tooling | $3,335 | $667 |
+| Raw Materials | $3,287 | $657 |
+| Electronics & Propulsion | $8,910 | $1,782 |
+| Consumables & Misc | $1,000 | $200 |
+| **TOTAL (5 UAVs, no labor)** | **$16,532** | |
+| **Per UAV (no labor)** | | **$3,306** |
+
+**Absolute minimum per UAV** (with budget optimizations from Section G): **~$2,616**
+
+## References
+
+1-20: See Draft 01 references (all still applicable)
+
+Additional sources:
+21. S-Glass vs E-Glass comparison: https://wiki-science.blog/s-glass-vs-e-glass-key-differences
+22. Reinforcement Fiber Reference: https://explorecomposites.com/materials-library/fiber-ref/
+23. S-Glass vs Carbon Fiber: https://carbonfiberfriend.com/s-glass-vs-carbon-fiber/
+24. RF Attenuation by composite materials: https://www.rocketryforum.com/threads/rf-attenuation-by-body-tube-nosecone.186634/
+25. Russian foamplast UAV (max radio transparency): https://bulgarianmilitary.com/2023/10/15/russia-unveils-foamplast-fpv-uav-with-max-radio-transparency/
+26. Albatross UAV Kit: https://store.appliedaeronautics.com/albatross-uav-kit/
+27. UAV spar connector development: https://www.konelson.net/home/spar-connector-development
+28. Scabro Innovations UAV prototyping: https://scabroinnovations.com/diensten/composite-airframe-prototyping/
+29. Tattu 330Wh/kg 6S pricing — GenStattu: https://genstattu.com/tattu-semi-solid-state-330wh-kg-33000mah-10c-22-2v-6s1p-g-tech-lipo-battery-pack-with-xt90-s-plug/
+30. Pixhawk 6X pricing — Holybro: https://holybro.com/products/pixhawk-6x-rev3
+31. RFD900x pricing — DrUAV: https://druav.com/products/rfdesign-rfd900x
+32. Composite workshop setup — Fibre Glast: https://www.fibreglast.com/blogs/learning-center/setting-up-a-composite-shop
+33. CNC hot wire foam cutter — Vortex-RC: https://www.vortex-rc.com/product/4-axis-diy-hot-wire-cnc-for-rc-hobbyists-aeromodellers-and-designers/
+34. Composite mold making — Canuck Engineering: https://www.canuckengineering.com/capabilities/composite-molds/
				`@@ -0,0 +1 @@`
				`I want to build a UAV plane for reconnaissance missions maximizing flight duration. Investigate what is the best frame material for that purpose`