Эксперимент: Tangent-Branch Neural Evasion Layer (#493)

{ "worked": true, "confidence": 8, "verdict": "Built a readable tangent-branch planner with analytic tangent points, hard/soft branch timing, and timestep-aware half-space velocity projection. The mechanism manifested: tangent distance and disk-clearance identities matched to 7.1e-16 and 2.2e-15 maximum error, the predicted symmetric branch switch occurred at 0 radians exactly, and the canonical speed-ratio threshold approached the predicted supremum of 2. Projection changed raw clearance from about -0.80 to nonnegative clearance in the residual sweep.", "metrics": { "baseline": "Unprojected residual field: minimum clearance -0.80; final position [-7.00, 0.00]. Residual sweep penetrations ranged approximately -0.20 to -0.80.", "idea": "Tangent/projection layer: tangent relative-distance error 7.06e-16, tangent disk-clearance error 2.22e-15, branch switch observed at 0.0 rad, speed-ratio supremum 2.0, projected sweep minimum clearance 0.00 to 0.04; mini-rollout minimum clearance 0.01 and final position [0.81, 0.00]." }, "how_to_run": "/home/maxwelhelp/main/bin/python3 tangent_evasion.py", "files": [ "tangent_evasion.py", "results.json", "run_output.txt" ], "limitations": "This is a deterministic geometric toy experiment, not PPO or multi-agent RL. It uses stationary disks and a simple Euler integrator; it does not test moving pursuers, learned residuals, training sample efficiency, path length or completion-time gains, GPU execution, model mismatch, or full two-pursuer interaction." }