After a deep investigation into the n=5 mecanum sim-to-real gap, all
attempted fixes (consensus tightening, wall_reject tightening, static-
phantom drop, deploy-time track merge, in-tracker track merge,
fp_rate-augmented retrain, max_range cap, 140° mecanum retrain) failed
to reliably pen n=5 in Webots without regressing n=10. The phantom
problem at 360° + small flock is genuinely hard and out of scope for
the deadline; documented in docs/status.md.
Result preserved from the previous mecanum work:
* 16/16 differential cells pen N/N.
* 4/8 mecanum cells (all n=10) pen 10/10 via Supervisor kinematic
injection (commit 27c0f65).
* n=5 mecanum is the known gap.
Small changes that survived the iteration:
* tests/test_config.py — strafe_efficiency=1.0 is now valid (kinematic
injection means the gym preset and Webots controller share the
formula, so textbook values produce gym-identical body motion).
* tools/run_webots.sh — refreshed the LiDAR-variant comment.
* training/rl/train.py — comment polish.
Replace the failing ODE-rolled mecanum chassis dynamics with a
Supervisor.setVelocity call that uses the gym mecanum forward
kinematics formula directly. Wheel motors still spin (visual);
chassis motion comes from the gym model so training and deployment
match by construction.
Results (seed=42, n=10 sheep): BC + RL mecanum pen 10/10 in both
field and field_round. n=5 mecanum cells still 0/5 due to tracker
phantoms anchored to wall corners under the 360° LiDAR — documented
in docs/status.md as the remaining gap.
Cleanup: drop deploy-time hacks (HERDING_HEADING_*, HERDING_OMEGA_CLAMP,
HERDING_TRACKER_*) that were workarounds for the old ODE chaos;
revert the proto inertiaMatrix, roller dampingConstant, and reduced
motor torque since they no longer carry load; refresh comments
around the mecanum config presets.
* training/bc/collect.py: --use-webots-preset now picks the
drive-matched variant. Mecanum drives get HERDING_MEC_WEBOTS
(with the Webots-calibrated strafe efficiency and bleed) so the
collected demos reflect the imperfect physical mecanum the
deployed policy will see. Differential drives still use
HERDING_WEBOTS (no behaviour change there).
* training/rl/train.py: mecanum fine-tune now *unconditionally*
applies the HERDING_MEC_WEBOTS robot config to the PPO env (the
policy must update against the same imperfect kinematics it
deploys on). Diff fine-tune unchanged.
To retrain a mecanum policy end-to-end against the new proto:
python -m training.bc.collect --drive-mode mecanum --world field \
--use-webots-preset \
--out training/bc/demos_mecanum_field_v2.npz
python -m training.bc.pretrain --demos training/bc/demos_mecanum_field_v2.npz \
--out training/runs/bc_mecanum_field_v2 ...
python -m training.rl.train --bc training/runs/bc_mecanum_field_v2 \
--out training/runs/rl_mecanum_field_v2 \
--drive-mode mecanum --world field --use-webots-preset
The same flow for field_round / mecanum/round.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Today's session worked across the full Webots delivery stack — found and
fixed a cluster of bugs blocking the BC/RL transfer, then explored
training-side mitigations for the residual perception gap.
Bug fixes:
- Makefile FP_RATE default 2.0 → 0.0: BC demos used fp_rate=0 but RL
fine-tune defaulted to fp_rate=2, poisoning the BC obs distribution
and stalling PPO at 0% success across 1.46M+ steps.
- controllers/{shepherd_dog,sheep}/runtime.ini: Webots was launching
controllers under system python3 (no numpy) and they were crashing
silently. Pinned to the conda tir env.
- herding/config.py HERDING_WEBOTS preset: pen_latch_depth 0.5 → 2.0,
max_new_tracks_per_step 3 → 1, static_reject 0.8 → 1.2. Stops phantom
FPs near the gate from latching as permanently-penned tracks.
- herding/perception/sheep_tracker.py: penned tracks now decay at
forget_steps × 8 instead of living forever. Adds get_positions
min_freshness filter for deploy-time use.
Training/eval matches deployment:
- training/bc/collect.py: --dagger-policy flag for DAgger rollouts
(policy drives, teacher labels) + --use-webots-preset for matched
140° tracker + DR config.
- controllers/shepherd_dog/shepherd_dog.py: scan-fallback (0, 0.6) when
BC/RL sees empty sheep_positions — recovers from FOV gaps.
Tooling:
- tools/dagger_round.sh: one-shot DAgger round (collect + concat + bc).
- tools/webots_sweep_gt.sh: full sweep with HERDING_USE_GT=1 for the
perception-gap diagnosis matrix.
- protos/ShepherdDog360.proto: 360° FOV variant for the FOV-ablation
comparison. Canonical proto stays at 140° per project spec.
Artifacts: v1 BC/RL policies for all 4 (drive × world) combos trained
in clean gym (success: diff/field 90-100%, diff/round 58%, mec/field
60-100%, mec/round 50-100%). DAgger r1/r2 BCs for diff/field show
12%→38% progression on gym HERDING_WEBOTS proxy but did not close
to actual Webots LiDAR (0/5 throughout). Next: LSTM policy or
learned tracker per the project-state memory.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Johnny Fernandes