Files
SkillFactory/configs/contact_cycle/simulation_functional_v1.json
chenlin 9fe9d27865 feat: checkpoint contact-cycle evaluation and RootXY PPO progress
Consolidate next-stage docs, archive superseded narratives, and preserve frozen document inputs. Retain invalid P2 pairs, partial RootXY training, failed attempts and nonselectable checkpoints as explicit limitations.

Include supervised evaluation/training, desktop GPU admission, bounded evidence and atomic-publication monitor repairs with CPU regressions. Ignore local dotenv configuration.

Validation: 1403 passed, 6 skipped for contact-cycle and related contracts/CLI tests. Full-suite legacy ABI and registration tests require Python 3.12/Torch 2.10/pxr; no runtime checks relaxed. No new physical run or accepted skill.
2026-10-15 14:18:17 +08:00

37 lines
2.3 KiB
JSON

{
"format": "skillfactory.contact-cycle-simulation-functional.v1",
"status": "simulation_functional_development_tuned_not_independent",
"scope": "One nominal 9s cycle of floating-root FL leg with declared active auxiliary fixture. Not full quadruped balance, learned-policy certification, robustness or sim2real.",
"physics_dt_s": 0.002,
"phase_durations_s": [3.0, 1.0, 0.6, 1.4, 0.8, 2.2],
"nominal_scenario_id": 0,
"gates": {
"minimum_complete_cycles": 1,
"support_final_window_s": 0.5,
"reload_final_window_s": 0.2,
"contact_force_on_N": 2.0,
"support_reload_contact_fraction_min": 0.8,
"support_reload_load_fraction_min": 0.5,
"unload_final_window_s": 0.1,
"unload_remaining_support_fraction_max": 0.8,
"swing_gap_m_min": 0.01,
"swing_air_fraction_min": 0.8,
"swing_contiguous_air_s_min": 0.1,
"swing_xy_progress_fraction_min": 0.5,
"landing_position_error_m_max": 0.05,
"root_height_error_m_max": 0.1,
"normal_force_negative_tolerance_N": 0.5,
"root_tilt_rad_max": 0.6,
"ground_penetration_m_max": 0.002,
"auxiliary_vertical_N_max": 140.0,
"auxiliary_lever_arm_m_max": 0.06,
"guide_vertical_N_max": 0.000001,
"normalized_effort_velocity_max": 1.00001,
"joint_limit_excess_rad_max": 0.01
},
"deferred": ["random disturbances", "cross-seed success confidence", "120s endurance", "sensor and actuator domain randomization", "hardware calibration", "sim2real"],
"separation_of_claims": "Report sim_functional_passed separately for analytic_reference and learned_policy. Keep historical strict results unchanged; accepted_skill remains false in this development checker. No implicit training grant.",
"development_note": "Initial prototype required unload<=0.5W; retrospective reference traces showed about0.676W with actual flight/landing. Function-first revision explicitly requires at least20% measured reduction relative to preceding support, not near-zero load tracking. Prototype failure/config are retained; this is development tuning, never independent acceptance.",
"runtime_note": "This file is an offline functional checker contract, not a change to the frozen task or its termination code. Partial historical episodes cannot be extrapolated past their actual stop. Any future relaxed runtime requires a new explicit protocol."
}