Skip to content

Latest commit

 

History

History
22 lines (13 loc) · 2.64 KB

robots.md

File metadata and controls

22 lines (13 loc) · 2.64 KB

Robots

To add a new robot, you can follow any of the existing robots built already as templates in ManiSkill. We also highly recommend that you read through the custom robot tutorial to learn how to make new robots and tune them.

ManiSkill is a supporter of open-source and we encourage you to make contributions to help grow our list of robots in simulation!

Contributing the Robot to ManiSkill

We recommend first opening an issue on our GitHub about your interest in adding a new robot as to not conflict with others and to consolidate information. Once done, our maintainers can give a go ahead.

In your pull request, we ask you to do the following:

  • The robot / agent class code should be placed in mani_skill/agents/<agent_group_name>/your_robot.py. If you want to re-use an agent class (e.g. as done with the Allegro hand robot and the Panda robot) you can create a folder that groups all the different agent classes together.
  • Update mani_skill/utils/download_asset.py and add a data source linking to your new robot under the initialize_extra_resources function. We recommend you to comment on the issue you opened with the asset files so we can then create a repository for it that is versioned. You can also create a repository yourself, you can follow how https://github.com/haosulab/ManiSkill-ANYmalC/ is setup to do this. The easiest way to allow access to assets is to create a Github release and link the source code zip file in the download_asset.py code
  • Finally document the new robot! Add an appropriate section to docs/source/robots and follow the template of the other robots there.

For some robot types we ask for additional verifications, currently of which we only ask for details for manipulation robots.

Testing Manipulation Robots

For any robot that is meant to be able to manipulate objects, we ask you to verify the robot can solve the PickCube-v1 task successfully via training with GPU simulation and RL (The PPO algorithm specifically). The PickCube-v1 task is a simple task where the robot learns to pick up a cube from a table and bring it to a green goal location. You will need to update the PickCube-v1 configs to make the task solvable with the new robot. The existing code shows examples of how they are already benchmarked for the Franka Emika robot, XArm6+Robotiq gripper Robot, and SO100 robot arm.