Skip to content

Dexterous grasping with RL in sparse reward environments

License

Notifications You must be signed in to change notification settings

amacati/dextgen

Repository files navigation

Dext-Gen: Dexterous Grasping in Sparse Reward Environments with Full Orientation Control

PEP8 Check Tests

In order to run the code from this project, install anaconda and create a conda environment from the provided environment file with

user@pc: $ conda create -f environment.yaml

, activate the new environment with

user@pc: $ conda activate dextgen

and install the dextgen packages from the project root folder with

user@pc: $ pip install .

Instructions on how to use the project are available in the package documentation. To read the documentation, build it from the docs directory with

user@pc: $ make html

and open the index file under 'docs/_build/html/index.html' in the browser of your choice, e.g. with

user@pc: $ firefox _build/html/index.html

The experiment environments are located in the envs module, the learning algorithm is located under mp_rl. The grasp optimization routines are organized in the optim module. Credit to code from external repositories is given in the individual source files.

If you have any issues with the installation of MuJoCo and mujoco-py such as the osmesa.h error, see the instructions on how to resolve common issues.

About

Dexterous grasping with RL in sparse reward environments

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages