Skip to content

Latest commit

 

History

History
76 lines (42 loc) · 2.72 KB

README.md

File metadata and controls

76 lines (42 loc) · 2.72 KB
Raw

Cloth Simulation Environment

This is the cloth simulation environment used in UniFolding. It is built on RFUniverse which is a a platform developed in Unity for robot simulation and reinforcement learning. The GPU-based cloth simulator in this project is based on Cloth Dynamics plugin in Unity Asset Store.

Requirements

Supported Operating System

  • Windows 11 with WSL2
  • Ubuntu 20.04 (recommended)

Setup

  1. Clone Project.

    git clone https://github.com/xiaoxiaoxh/UniFolding

  2. Install Unity Hub and install Unity Editor 2022.3 LTS through Unity Hub.

    Install the Unity Hub | Unity Hub | 3.0

    You can install it according to the official documentation:

    Downloading and installing Editors and modules with the Unity Hub | Unity Hub | 3.0

  3. Add and Open Project (RFUniverse) in Unity Editor.

  4. The ClothDynamics plugin is used in the project for clothing simulation. Since it is a paid plug-in, you need to purchase it and import it into the project.

    Unity Asset Store Link：Cloth Dynamics | Unity Asset Store

    The version that has been verified and used in the project is 2.0.9 beta.

    When importing, do not replace existing scripts with the same name in the project. Because there are some modifications.

    image

Minimal Example

  1. Open And Play RFUniverse/Assets/Scenes/Scene.unity

    image

  2. Select the Mainobject in the scene

    image

  3. Click the LoadMesh button to load clothes.

    Click the RandomFold button to random folding.

    image

Preview:

image

Build executable program

Follow the normal process to build an executable program. You can follow the official documentation:

Unity - Manual: Publishing Builds

This executable program supports two-way messaging with the python interface (pyrfuniverse).

Customization

Currently, we have implemented some basic primitives (e.g. fling) in Main.cs. If you want to implement your own action primitives, please refer to the documentation of dynamic message.