Progressive Learning for Physics-informed Neural Motion Planning
Ruiqi Ni, Ahmed H Qureshi

Paper | GitHub | arXiv | Published in RSS 2023.

This repository is the official implementation of "Progressive Learning for Physics-informed Neural Motion Planning".

Clone the repository into your local machine:

git clone https://github.com/ruiqini/P-NTFields --recursive

Install requirements:

conda env create -f NTFields_env.yml
conda activate NTFields

Download datasets and pretrained models, exact and put datasets/ Experiments/ to the repository directory:

Datasets and pretrained model

The repository directory should look like this:

P-NTFields/
├── datasets/
│   ├── arm/    # 6-DOF robot arm, cabinet environment
│   ├── c3d/    # C3D environment
│   ├── gibson/ # Gibson environment
│   └── test/   # box and bunny environment
├── Experiments
│   ├── UR5/   # pretrained model for 6-DOF arm
│   └── Gib_multi/    # pretrained model for Gibson
•   •   •
•   •   •

To prepare the Gibson data, run:

python dataprocessing/preprocess.py --config configs/gibson.txt

To prepare the arm data, run:

python dataprocessing/preprocess.py --config configs/arm.txt

To visualize our path in a Gibson environment, run:

python test/gib_plan.py 

To visualize our path in the 6-DOF arm environment, run:

python test/arm_plan.py 

To train our model in multiple Gibson environment, run:

python train/train_gib_multi.py

To train our model in the 6-DOF arm environment, run:

python train/train_arm.py 

Example 1

Example 2

Example 3

Please cite our paper if you find it useful in your research:

@article{ni2023progressive,
  title={Progressive Learning for Physics-informed Neural Motion Planning},
  author={Ni, Ruiqi and Qureshi, Ahmed H},
  journal={arXiv preprint arXiv:2306.00616},
  year={2023}
}

This implementation takes EikoNet and NDF as references. We thank the authors for their excellent work.

P-NTFields is released under the MIT License. See the LICENSE file for more details.