Skip to content

fix(phase3): harden LQR allocation, provably collision-free A*, dynamic obstacle avoidance, physics integration, and complete provenance - #16

Merged
AryanXCode646 merged 7 commits into
feat/arl-autonomous-navigationfrom
fix/phase3-correctness-gaps
Nov 16, 2026
Merged

AryanXCode646 merged 7 commits into
feat/arl-autonomous-navigationfrom
fix/phase3-correctness-gaps

Conversation

@AryanXCode646

@AryanXCode646 AryanXCode646 commented Nov 15, 2026 •

Copy link
Copy Markdown
Owner

Summary of Engineering Hardening & Correctness Closure (Phase III.2 Adversarial Safety, Collision & Provenance Verification)

P0 — Adversarial Dynamic Collision & Boundary Reflection Decomposition

  • Exact Boundary Reflection Decomposition: Implemented DynamicObstacleSphere3D.get_bounce_times(t0, t1) to analytically compute all exact boundary reflection timestamps. Splitting the continuous trajectory $[t_0, t_1]$ at reflection events guarantees piecewise linear motion, completely eliminating false negatives near boundary walls where chords cut across the arena interior.
  • Defensible Vehicle Curvature Envelopes: In _check_subsegment_collision, intermediate trajectory absence evaluates against the full step duration $\varepsilon_{\text{veh}} = \frac{1}{8} a_{\max} (\Delta t)^2$, preventing the unvalidated $num_substeps^2$ bound shrinkage flaw. When genuine RK4 substeps are provided, each substep is safely bounded by $\frac{1}{8} a_{\max} (\Delta \tau)^2$.
  • Strict Parameter Safety & Reversal Rejection: Rejects reversed time intervals ($t_1 < t_0$) with ValueError, fixing the previous silent acceptance under $dt \le 10^{-9}$. Validates finite non-negative prev_time, curr_time, vehicle_radius, veh_acc_max, and obstacle parameters. Zero-duration checks verify both start and end positions.

P0 — Defensible Operating Envelope & Analytical Acceleration Bounds

  • Operating Envelope Enforcement: DroneDynamics6DOF.max_translational_acceleration() actively monitors vehicle state velocity and recent RK4 substeps (last_substep_velocities). Ensures the evaluated operating speed envelope satisfies $v_{\text{envelope}} = \max(v_{\text{bound}}, v_{\text{actual}})$, guaranteeing acceleration bounds never underestimate high-speed off-nominal flight.
  • Analytical Terminal Velocity: When max_speed_bound is omitted (None), computes exact terminal velocity $v_{\text{term}}$ under continuous maximum thrust, gravity, and quadratic drag. Fails closed if physical parameters (mass, gravity, thrust, damping, wind) are invalid or non-finite.
  • Elimination of Arbitrary Empirical Fallbacks: Removed arbitrary constant fallbacks (else 35.0) from FlightScenarioRunner.run_scenario(); fails closed with ValueError if vehicle dynamics cannot provide a defensible acceleration bound.

P1 — Deep Recursive Callable Provenance & Descriptor Introspection

  • Descriptor Unwrapping Across MRO: Unwraps and fingerprints staticmethod (.__func__), classmethod (.__func__), and property (.fget, .fset, .fdel) descriptors in custom classes and callable instances. Modifying any descriptor method alters the class and callable fingerprint deterministically.
  • Nested Module Dependency Resolution: Bytecode attribute chains (pkg.submod.helper) referenced in co_names are recursively resolved and fingerprinted down to underlying functions and constants.
  • Lexical Cycle References: Self-recursive and circular callable dependencies terminate cleanly via lexical references (recursion_ref) without infinite recursion.
  • Fails-Closed Incomplete Tracking: Unsupported opaque objects and dynamic types safely mark provenance_complete = False and record structured diagnostic reasons without leaking unstable memory addresses.

P2 — Documentation & Scientific Qualification

  • Updated docs/AUTONOMOUS_NAVIGATION.md with:
    • Analytical operating speed envelope and terminal velocity derivations.
    • Exact boundary reflection decomposition mechanics.
    • Substep vs full-interval curvature error bounds.
    • Explicitly qualified conditional safety contracts (stating assumptions on grid resolution $\Delta$, vehicle clearance margin $\delta_{\text{margin}}$, and state estimation bounds).

Local Verification

  • Pytest: 577 passed, 0 failed, 3 warnings in 85.18s (tests/, 100% pass rate).
  • Ruff Linter: 0 errors (ruff check .).
  • Ruff Formatter: 118 files clean (ruff format --check .).
  • Mypy Typechecker: 0 errors across 58 source files (mypy src/).

GitHub Actions CI Verification

All 10 checks passed on commit f703f66cf32eaddcd9774e066cc87b4ad3605a33:

  • CI / Lint & Typecheck (pull_request): SUCCESS (2m 59s)
  • CI / Lint & Typecheck (push): SUCCESS (2m 48s)
  • CI / Test Python 3.10 (pull_request): SUCCESS (3m 14s)
  • CI / Test Python 3.10 (push): SUCCESS (3m 43s)
  • CI / Test Python 3.11 (pull_request): SUCCESS (3m 09s)
  • CI / Test Python 3.11 (push): SUCCESS (3m 12s)
  • CI / Test Python 3.12 (pull_request): SUCCESS (4m 04s)
  • CI / Test Python 3.12 (push): SUCCESS (4m 06s)
  • CI / End-to-End Pipeline Smoke Test (pull_request): SUCCESS (1m 43s)
  • CI / End-to-End Pipeline Smoke Test (push): SUCCESS (1m 54s)

Workflow Runs:

…ic obstacle avoidance, physics integration, and complete provenance
…anning, and provenance

- Eliminate mutable dynamics state aliasing and calculate genuine inertial acceleration in IMU telemetry
- Enforce authoritative mission completion requiring sequential waypoint traversal, position settling, velocity settling, and collision freedom
- Preserve exact finite IEEE 754 precision in canonical JSON dumps and comprehensive scenario provenance hashing
- Implement continuous swept-sphere dynamic obstacle collision detection with adaptive temporal subintervals
- Implement closed-loop state-estimation-in-the-loop control contract and dedicated scenario
- Enable arbitrary non-grid-aligned 3D A* start/goal coordinates with collision-free neighbor connection and validation
- Couple stage-wise continuous first-order actuator lag across RK4 evaluation stages eliminating integration bias
- Add full test suites covering all residual defect requirements with 528 passing tests
… dynamic collision, complete provenance inference, and branch CI
…rovenance, and enforce provenance consistency
…rena bounds, complete recursive provenance, and verify 6-DOF continuous collision
…d static/boundary safety, and recursive provenance
…ng, defensible acceleration, and recursive descriptor provenance
@AryanXCode646
AryanXCode646 merged commit bde8c37 into feat/arl-autonomous-navigation Nov 16, 2026
10 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant