Skip to content

Latest commit

 

History

History
executable file
·
41 lines (30 loc) · 2.13 KB

README.md

File metadata and controls

executable file
·
41 lines (30 loc) · 2.13 KB
Raw

uf-gym

Adding UFACTORY robot model support to panda-gym training and simulation tools for Reinforcement Learning (RL). This repository is developed based on the original panda-gym framework and should work alongside with it.

Please check the panda-gym page for detailed documentation and credit information.

Prerequisites

  1. Python 3.8+ and pytorch;
  2. Nvidia driver and CUDA with proper versions for your GPU device.
  3. stable baselines 3 and extended module sb3-contrib.
  4. pybullet
  5. panda-gym
  6. Other dependent packages needed during installation or running process.

Usage

  1. The scripts have self-explanatory names, to train xArm7 with pick and place task:
$ python train_xarm7_pickNplace.py
  1. To test the trained model for pick and place task:
$ python test_xarm7_pickNplace.py

The algorithm, hyper-parameters, task, model name, saving directories etc. can all be modified in the code according to your needs.

Currently only xArm6 and xArm7 (with gripper) are supported, other models will be added in the future.

Training Results Demo
XArm6Reach-v3 XArm6Push-v3
XArm6Reach-v3 XArm6Push-v3
XArm7Slide-v3 XArm7PickAndPlace-v3
XArm7Slide-v3 XArm7PickAndPlace-v3

Pre-trained models

Some pre-trained models are uploaded and available on Hugging Face platform.