dora-mujoco simulation node with example for controlling any arm #1012
dora-mujoco simulation node with example for controlling any arm #1012haixuanTao merged 15 commits intodora-rs:mainfrom
Conversation
|
I think it's a good idea to store the meshes and exposes the xml. Making things slightly easier. I probably missed it but how do you make sure that the file are going to be retrieved by mujoco? |
Currently the meshes are downloaded using huggungface repo, the mesh files are automatically discovered through MuJoCo's XML-based model definition system, you can see that inside the panda.xml file Sorry for the late reply |
|
Isn't the urdf loaded from https://github.com/robot-descriptions/robot_descriptions.py?tab=readme-ov-file? |
|
Can we rename |
|
Can we also make Erika franks an example of dora-mujoco instead of a standalone node? |
|
Yes the mujoco sim node uses that, but mujoco franka node still uses hugging face to get the assets, also i have modifed them a bit to include the table ans the small cube Should have made a separate PR for that 😅 |
Ok will do that |
There was a problem hiding this comment.
It seems that this controller could work on any arm with an urdf right?
Maybe we could naming more apparent or directly use existing dora-pytorch-kinematics?
I wish this tutorial to have as many reusable component as possible.
There was a problem hiding this comment.
yes the controller should work with ant urdf, will try to make it reusable for any arm.
| self.collision_influence_distance = 0.05 | ||
|
|
||
| # Define which links can collide with each other (non-adjacent only) | ||
| self.collision_pairs = [ |
There was a problem hiding this comment.
so it seems that every pair can collide with each other?
Can't we have a more automated way to compute this?
There was a problem hiding this comment.
I had tried using PyTorch Volumetric to get this. It was having too much wobbly motion. And the collision avoidance part is not very reusable for any other arm
This pull request introduces the initial setup for the
dora-franka-mujocoproject, including documentation, configuration files, and assets for simulating the Franka Emika Panda robot in MuJoCo. The most important changes include Defining a YAML configuration for nodes, adding robot assets and MuJoCo XML files, and creating a Python package structure for the project.Documentation and Configuration:
README.mdfile with installation, formatting, linting, and testing instructions for thedora-franka-mujocoproject.demo.ymlto define the configuration for nodes, includinggamepadandmujoco_franka, with their inputs, outputs, and dependencies.Robot Simulation Assets:
panda.xmlto define the MuJoCo model for the Franka Emika Panda robot, including its joints, actuators, collision meshes, and visual meshes.scene.xmlto define the simulation environment, including a table, a movable box, and visual settings for the scene.names.txt) and corresponding 3D model files for the robot's components, such as links, fingers, and the hand.