This library implements Stochastic Gaussian Process Motion Planning algorithm in PyTorch. In Imitation Learning, we use this planner to sample trajectories from the learned Energy-Based Models encoding expert trajectories.
(Note: Planning code has been refactored from the original source repository.)
Activate your Python/Conda environment and install
pip install -e .
Additionally, please install https://github.com/anindex/torch_robotics.
Try planning in planar environment with multiple goals
python examples/planar_environment.py
Try planning in Panda environment with multiple obstacles
python examples/panda_environment_stochgpmp.py
If you encounter the exception regarding Positive-Definite matrix while initializing MP Priors, try to use higher precision floating point, e.g. torch.float64.
[1] Urain, J.* ; Le, A.T.* ; Lambert, A.*; Chalvatzaki, G.; Boots, B.; Peters, J. (2022). Learning Implicit Priors for Motion Optimization, IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS).
If you found this work useful, please consider cite us as below :-)
@inproceedings{iros2022_ebmtrajopt,
author = "Urain, J. and Le, A.T. and Lambert, A. and Chalvatzaki, G. and Boots, B. and Peters, J.",
year = "2022",
title = "Learning Implicit Priors for Motion Optimization",
booktitle = "IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)",
key = "motion planning, energy-based models",
URL = "https://www.ias.informatik.tu-darmstadt.de/uploads/Team/AnThaiLe/iros2022_ebmtrajopt.pdf",
crossref = "p11531"
}
If you have any questions or find any bugs, please let me know: An Le an[at]robot-learning[dot]de