huggingface
The inverse kinematic solution code of so-100
Are there any code of inverse kinematic of so-100, which just need the input of the x, y on my desk, then it can move to the target coordinate？ Thanks for any response.
We were discussing it, but I think that nobody started developing it. It should be pretty - just like 2/3 lines of code.
The only important note is that during the building leader arm for SO-100 we remove on gear, so the instruction should adapted to have inverse kinematics.
I second this, supporting a traditional control method would be superuseful. I tried but could not make it work with my arm.
Hi! I noticed this issue and wanted to share that I’ve implemented inverse kinematics support, including:
- Forward kinematics via DH parameters.
- Analytical Jacobian computation.
- Inverse kinematics using Damped Least Squares (DLS) to avoid singularities.
- Pose interpolation: linear (position) + SLERP (orientation)
Submitted an upgrade here: https://github.com/huggingface/lerobot/pull/1001 🚀
Hi! I noticed this issue and wanted to share that I’ve implemented inverse kinematics support, including:
- Forward kinematics via DH parameters.
- Analytical Jacobian computation.
- Inverse kinematics using Damped Least Squares (DLS) to avoid singularities.
- Pose interpolation: linear (position) + SLERP (orientation)
Submitted an upgrade here: #1001 🚀
Hey, just a note, I think you accidentally Closed the PR instead of leaving it open-
I think phospho.ai made some nice open source implementations, you can check its documentation https://docs.phospho.ai/learn/kinematics
