# YanjieZe/3D-Diffusion-Policy

[RSS 2024] 3D Diffusion Policy: Generalizable Visuomotor Policy Learning via Simple 3D Representations

Repository: https://github.com/YanjieZe/3D-Diffusion-Policy
Canonical: https://ross.abutalabs.com/products/3d-diffusion-policy
Homepage: https://3d-diffusion-policy.github.io
Language: Python
License: MIT
License Family: permissive
Last push: 2025-10-17T05:55:25+00:00

## Health v2 (maintenance only)
Score: 46/100 (v2, computed 2026-09-03T02:39:23.370411+00:00)
- activity 47, release rhythm 35, longevity 65
- inputs: {"age_days": 911, "days_push": 320, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1429, forks 168 (observed 2026-08-28T04:04:42.117313+00:00)

## What it is
3D Diffusion Policy (DP3) is a visual imitation learning algorithm that combines compact 3D point cloud representations with diffusion policies for robot manipulation. It is the official PyTorch research codebase from an RSS 2024 paper, supporting training and evaluation on simulated and real-robot tasks.

## Use cases
- train visuomotor policies from a handful of demonstrations
- learn robot manipulation skills from point cloud observations
- benchmark imitation learning baselines on simulated manipulation tasks
- deploy diffusion policies on real robot arms
- research generalizable robot policy learning with 3D representations

## When to choose
- you need sample-efficient imitation learning for robot manipulation
- your robot observations include point clouds or depth cameras
- you want a strong, well-cited baseline for visuomotor policy research
- you have few human demonstrations (10-40) per task

## When to avoid
- you need a production-ready robotics framework rather than research code
- your setup is RGB-only with no depth or point cloud data
- you lack a GPU or robotics simulation environment
- you want plug-and-play deployment without ML expertise

## Facets
- artifact type: library
- maturity: active
- function: machine-learning, deep-learning, robotics, simulation
- domain: robotics, machine-learning, deep-learning, artificial-intelligence
- platform: python
- tags: diffusion-policy, imitation-learning, point-clouds, visuomotor, robot-manipulation, research-code, linux, gpu

## Member repositories
- YanjieZe/3D-Diffusion-Policy (main) score 46

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:04:42.117313+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T04:37:14.399271+00:00, confidence not recorded.
  - readme: https://github.com/YanjieZe/3D-Diffusion-Policy (fetched 2026-08-28T04:04:42.117313+00:00, sha fc55785aa353)
  - homepage: https://3d-diffusion-policy.github.io (fetched 2026-08-29T11:48:48.881019+00:00, sha 63cf7926dd79)
- Data as of 2026-08-30T08:39:29.467469+00:00.