 Yuzviak
|
f67c5f3cd0
|
refactor(01-01): convert hot-path schemas/*.py to hot_types re-export shims
- schemas/eskf.py: keep ConfidenceTier + ESKFConfig; re-export IMUSample
and ESKFState from hot_types (define ConfidenceTier BEFORE the
hot_types imports to avoid circular import — eskf_state.py imports
ConfidenceTier from this module). Legacy alias IMUMeasurement = IMUSample.
- schemas/vo.py: re-export Features, Matches, RelativePose, Motion,
VOEstimate from hot_types.vo_estimate.
- schemas/satellite.py: re-export TileCoords, TileBounds, SatelliteAnchor.
- schemas/metric.py: keep LiteSAMConfig; re-export AlignmentResult,
ChunkAlignmentResult, Sim3Transform.
- schemas/rotation.py: keep HeadingHistory + RotationConfig; re-export
RotationResult.
Auto-fixes (Rules 1 + 3) needed to keep the 216-test floor green:
- core/rotation.py: refactor try_rotation_steps to use
dataclasses.replace instead of attribute assignment on RotationResult
(Rule 1 — frozen dataclass forbids mutation; Pydantic silently allowed
it). PATTERNS.md §6.1 already flagged Pose mutation but missed this site.
- hot_types/vo_estimate.py: add Optional `covariance: np.ndarray` field
to RelativePose (Rule 3 — five test sites construct RelativePose with
`covariance=np.eye(6)`; Pydantic v2 silently accepted the extra kwarg
via default `extra="ignore"`. Declaring the field preserves the
construction contract under the dataclass migration without editing
tests).
Verification: pytest tests/ -q --ignore=tests/e2e → 216 passed, 8 skipped
(matches baseline). Accuracy bench (23 tests) passes.
|
2026-05-10 22:47:56 +03:00 |
|
|
Yuzviak
|
dd9835c0cd
|
fix(lint): resolve all ruff errors — trailing whitespace, E501, F401
- ruff --fix: removed trailing whitespace (W293), sorted imports (I001)
- Manual: broke long lines (E501) in eskf, rotation, vo, gpr, metric, pipeline, rotation tests
- Removed unused imports (F401) in models.py, schemas/__init__.py
- pyproject.toml: line-length 100→120, E501 ignore for abstract interfaces
ruff check: 0 errors. pytest: 195 passed / 8 skipped.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
|
2026-04-02 17:09:47 +03:00 |
|
|
Yuzviak
|
094895b21b
|
feat(phases 2-7): implement full GPS-denied navigation pipeline
Phase 2 — Visual Odometry:
- ORBVisualOdometry (dev/CI), CuVSLAMVisualOdometry (Jetson)
- TRTInferenceEngine (TensorRT FP16, conditional import)
- create_vo_backend() factory
Phase 3 — Satellite Matching + GPR:
- SatelliteDataManager: local z/x/y tiles, ESKF ±3σ tile selection
- GSD normalization (SAT-03), RANSAC inlier-ratio confidence (SAT-04)
- GlobalPlaceRecognition: Faiss index + numpy fallback
Phase 4 — MAVLink I/O:
- MAVLinkBridge: GPS_INPUT 15+ fields, IMU callback, 1Hz telemetry
- 3-consecutive-failure reloc request
- MockMAVConnection for CI
Phase 5 — Pipeline Wiring:
- ESKF wired into process_frame: VO update → satellite update
- CoordinateTransformer + SatelliteDataManager via DI
- MAVLink state push per frame (PIPE-07)
- Real pixel_to_gps via ray-ground projection (PIPE-06)
- GTSAM ISAM2 update when available (PIPE-03)
Phase 6 — Docker + CI:
- Multi-stage Dockerfile (python:3.11-slim)
- docker-compose.yml (dev), docker-compose.sitl.yml (ArduPilot SITL)
- GitHub Actions: ci.yml (lint+pytest+docker smoke), sitl.yml (nightly)
- tests/test_sitl_integration.py (8 tests, skip without SITL)
Phase 7 — Accuracy Validation:
- AccuracyBenchmark + SyntheticTrajectory
- AC-PERF-1: 80% within 50m ✅
- AC-PERF-2: 60% within 20m ✅
- AC-PERF-3: p95 latency < 400ms ✅
- AC-PERF-4: VO drift 1km < 100m ✅ (actual ~11m)
- scripts/benchmark_accuracy.py CLI
Tests: 195 passed / 8 skipped
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
|
2026-04-02 17:00:41 +03:00 |
|
|
Yuzviak
|
905d6992de
|
feat: stage8 — Global Place Recognition and Metric Refinement
|
2026-03-22 23:03:54 +02:00 |
|