Piper Diffusion Policy — Hanyang
Diffusion Policy for a Piper single-arm robot, trained on the Hanyang 100-demo normalized pick-and-place dataset.
Model details
| Field | Value |
|---|
| Policy | Diffusion Policy |
| Robot | Piper |
| Dataset | Hanyang 100-demo normalized dataset |
| Training steps | 100,000 |
| Scheduler | DDIM |
| Inference steps | 20 |
| Cameras | Front, right camera |
Usage
Use this repository as the policy path in a LeRobot inference or rollout command:
--policy.path=leejaehot/piper-dp-hanyang-v1This repository contains the model weights and policy processors. Training-state files are not included.
