2 Commits

Author	SHA1	Message	Date
Victor Mylle	8cc84d6a21	feat: RMA-style history-conditioned policy for sim2real adaptation ... Added a temporal observation history buffer and 1D-CNN encoder so the policy can implicitly infer environment parameters (mass, friction, gear ratios, etc.) from recent (obs, action) dynamics. Architecture: history window [(obs₀,a₀), ..., (obs_{H-1},a_{H-1})] → 1D-CNN HistoryEncoder → embedding (32-dim) → concat [current_obs, embedding] → MLP → action Components: - BaseRunner: history ring buffer, _push_history/_reset_history, augmented obs space (6 + H×7 = 76 with H=10) - HistoryEncoder (src/models/mlp.py): 2-layer temporal Conv1d + GAP - SharedMLP: optional history_length/raw_obs_dim/embedding_dim params; splits augmented obs, encodes history, feeds [obs, emb] to MLP - TrainerConfig: history_length, embedding_dim fields - All runner configs: history_length=10 by default - Tests: encoder shape, model with/without history, config defaults	2026-03-28 18:58:24 +01:00
Victor Mylle	5880997786	refactor: merge motor sysid into unified sysid module ... Unified the two separate sysid codepaths (motor-only and full-system) into a single module that optimizes all 28 parameters jointly: - 13 motor params (asymmetric gear, damping, friction, deadzone, Stribeck boost, action bias, filter tau, armature, ctrl_limit) - 15 pendulum/arm params (mass, CoM, inertia, joint dynamics) Key changes: - Added stribeck_friction_boost, stribeck_vel, action_bias to ActuatorConfig (robot.py) and MJX runner - Created shared src/sysid/preprocess.py (SG velocity recomputation) - Rewrote src/sysid/rollout.py with unified MOTOR_PARAMS + PENDULUM_PARAMS spec and PARAM_SETS dict for flexible subset optimization - Updated optimize.py, export.py, visualize.py to use unified params (removed all LOCKED_MOTOR_PARAMS references) - Removed src/sysid/motor/ module and scripts/motor_sysid.py Net: -1383 lines, zero code duplication between motor and full-system sysid.	2026-03-28 16:48:22 +01:00