omkarpatil/push-tape-right-dp-wrist-diffusion

Diffusion Policy — push-tape-right (wrist cameras only)

LeRobot Diffusion Policy trained on push-tape-right for the ROBOTIS FFW SG2 Rev1, using shared-norm statistics pooled across its composition group.

Cameras

cam_left_wrist, cam_right_wrist — native resolution (both wrist cameras are 424x240).

Diffusion Policy requires every camera to share one resolution, and this robot's do not (head 376x672, wrist 424x240). The wrist-only variant is uniform as recorded; the 3-camera variant required re-encoding all views to a common size.

Composition group A

Members: push-tape-left, push-tape-right. Normalization statistics were pooled over 5,768 frames of all group members and written identically into each member's dataset:

sha256(observation.state + action, normalization fields)[:12] = 839f172565ff

Only compose models reporting this same hash.

Note on cross-architecture composition: the GR00T policies for these same tasks share the same pooled statistics file but consume different fields — GR00T uses q01/q99 min-max (use_percentiles=True), Diffusion Policy uses min/max (STATE/ACTION default to MIN_MAX). So diffusion composes with diffusion and GR00T with GR00T, but not across the two.

Training — LeRobot defaults

Policy	diffusion (LeRobot 0.6.1, ROBOTIS lerobot-cyclo fork)
Steps	100 000
Batch size	8
Optimizer	lr 1e-4, betas (0.95, 0.999), weight decay 1e-6
Noise scheduler	DDPM
Params	274472400
Data rate	15 fps
Final train loss	0.003

Dataset in LeRobot v3.0 format (converted from v2.1, with pooled statistics restored after conversion — the v2.1→v3.0 converter regenerates stats and would otherwise have replaced the group-pooled values with per-task ones).