Skip to content
RobothecaBeta
Log in

Diffusion Policy

  • Software
  • Humanoids and robot learning
  • Framework

A learning framework that trains robot visuomotor policies by generating actions with a diffusion model, with reproduced benchmarks in simulation and a real UR5 robot setup.

  • Stale

Made by Columbia AI and Robotics Lab, now the Stanford REAL lab, United States (checked on github.com/real-stanford (opens in a new tab)).

Source on GitHub: github.com/real-stanford/diffusion_policy (opens in a new tab). Facts read at commit 5ba07ac (opens in a new tab).

Licenses

Code
MIT (opens in a new tab)

Read by Robotheca from LICENSE (opens in a new tab) (MIT) at commit 5ba07ac (opens in a new tab), on 16 Oct 2026.

Commercial use
Yes

Labels

  • StaleNo commit in 12 months: the newest is from 24 Dec 2024.

What it does

  • Trains robot control policies from demonstrations using a diffusion model that produces action sequences.
  • Provides configs, logs and checkpoints for the experiments behind the paper's benchmark tables.
  • Includes evaluation scripts for pre-trained checkpoints on simulated tasks such as Push-T.
  • Supports both state-based and image-based observations for low-dimensional and vision tasks.
  • Offers demonstration collection, training and evaluation workflows for a real robot using a SpaceMouse.
  • Includes a modular codebase structure for adding new tasks and policy methods.

Written by Robotheca from the project's README (opens in a new tab) at commit 5ba07ac.

Getting started

Difficulty
Intermediate
Hardware it needs, as the project states it
For simulation, a Linux machine with an Nvidia GPU; for the real robot, a UR5-CB3 or UR5e arm, two RealSense D415 cameras, a 3Dconnexion SpaceMouse, and Ubuntu 20.04.3 (tested).

Maintenance

Status
Stale: no commit in 12 months
Newest commit
24 Dec 2024 on main
Latest release
None published on GitHub

Checked on GitHub on 16 Oct 2026.

Links

Robots on Robotheca that use it

No robot on Robotheca is linked to it yet. Links come from a robot's robot.yaml, from its dependency files read as text, or from Robotheca's editors.

Robotheca lists software as a reference: nothing is hosted or run here, and every fact links to where it was read.