SE(3) Linear Parameter Varying Dynamical Systems for Globally Asymptotically Stable End-Effector Control
Implementation of SE(3) Linear Parameter Varying Dynamical Systems (LPVDS) for pose control. The main branch contains the coupled learning of pose trajectory, whereas the decoupled variant of this framework is maintained in the main-decoupled branch, and please refer to its README.md for usage.
The available dataset contains the following:
- clfd_data: collection of pose trajectory from the clfd framework
- kinesthetic_demo: pose trajectory collected from kinesthetic teaching
- UMI_demo: single pose trajectory collected from the UMI gripper real-time data collection
Create a virtual envrionemnt and install the required packages:
virtualenv -p /path/to/python3.8 venv
source venv/bin/activate
pip install -r requirements.txt
Begin the learning process and evaluate the result:
python main.py