Motion Planning algorithms

Install

git clone git@github.com:Obat2343/READ.git
mkdir git
cd git

Install followings in the git directory.

• Pyrep (https://github.com/stepjam/PyRep)
• CoppeliaSim (https://www.coppeliarobotics.com/previousVersions) # Please check the Pyrep repository to confirm the version of CoppeliaSim
• RLBench (https://github.com/Obat2343/RLBench)
• RotationConinuity (https://github.com/papagina/RotationContinuity)

Prepare Dataset

To create the dataset for training and testing, please run the following command.

python create_dataset.py --task_list TaskA TaskB

---

For TTI-IIM students

mkdir dataset
cd dataset
ln -s /misc/dl001/dataset/ooba/RLBench4 .
ln -s /misc/dl001/dataset/ooba/RLBench4-panda RLBench-test

Train

Please train a VAE first.

cd main
python READ/Train_ACTOR.py --config_file ../config/Train_ACTOR.yaml

Then, train a diffusion model.

cd main
python READ/Train_READ.py --config_file ../config/Train_READ.yaml

Test

cd main
python READ/Evaluate_READ.py --config_file ../config/yamlfilename --model_path /path/to/pthfile --tasks PickUpCup --inf_method_list retrieve_from_SPE

in server:

xvfb-run python READ/Evaluate_READ.py --config_path ../config/user/Test_LatentContinuousDiff.yaml --model_path /path/to/pthfile --tasks PickUpCup --inf_method_list retrieve_from_SPE --off_screen

---

@inproceedings{oba2024read,
  title={READ: Retrieval-Enhanced Asymmetric Diffusion for Motion Planning},
  author={Oba, Takeru and Walter, Matthew and Ukita, Norimichi},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  pages={17974--17984},
  year={2024}
}