kinder-diffusion-policy

A fork of diffusion_policy adapted for the KinDER.

---

Overview

kinder-diffusion-policy extends upstream diffusion_policy with the following capabilities:

Training

• Support different config files for both 2D and 3D environments in KinDER.
• Support mobile manipulator policy in KinDER

Inference

• Provide a policy server to evaluate different environments in KinDER

---

Installation

Clone the repository:

git clone git@github.com:Princeton-Robot-Planning-and-Learning/kinder-diffusion-policy.git

---

Environment Setup

$ mamba env create -f conda_environment.yaml

or

$ conda env create -f conda_environment.yaml

---

Training

Step 1: Prepare Data

Follow the README for KinDER Imitation Learning Baselines

---

Step 2: Launch Training

cd ~/kinder-diffusion-policy
mamba activate robodiff
python train.py --config-name=train_sweep3d_image

---

Evaluation

Step 1: Run the evaluation server

cd ~/kinder-diffusion-policy
mamba activate robodiff
python policy_server.py --ckpt-path $Checkpoint_path 

Step 2: Launch the KinDER environment

Follow the README for KinDER