ManiSkill Robots

This package provides a set of robots for the ManiSkill environment. It includes various robot configurations and their corresponding URDF files.

1. Prepare Environment

Option 1: use uv as environment manager

• create a new environment

  uv venv

• install pytorch (Use NJU mirror)

  uv pip install torch torchvision torchaudio --index-url https://mirrors.nju.edu.cn/pytorch/whl/cu128

• check torch cuda enable

  python
  import torch 
  torch.cuda.is_available()

• install mani_skill (Use NJU mirror)

  uv pip install --upgrade mani_skill --prerelease=allow --index-url https://mirrors.nju.edu.cn/pypi/web/simple

2. Usage

2.1 Only load robot

• Visualize the robot

  python scripts/demo_robot.py --shader="rt-fast" -r "arx_r5"

• Random action

  python scripts/demo_robot.py -c "pd_joint_delta_pos" --random-actions -r "dobot_cr5" --shader="rt-fast"

  # GPU Simulation
  python scripts/demo_robot.py  -c "pd_joint_delta_pos" -b "gpu" --random-actions -r "arx_r5" --shader="rt-fast"

2.2 Load Robot with Environment

• Visualize the robot in the environment

  python scripts/demo_random_action.py -e "PushCube-v2" --shader="rt-fast" --render-mode="human" -r "arx_r5"

  python scripts/demo_random_action.py --shader="rt-fast" -e "PickCube-v2" --render-mode="human" -r "piper"