HOPE: A Reinforcement Learning-based Hybrid Policy Path Planner for Diverse Parking Scenarios

This repository contains code for the paper HOPE: A Reinforcement Learning-based Hybrid Policy Path Planner for Diverse Parking Scenarios. This work proposes a novel solution to the path-planning task in parking scenarios. The planner integrates a reinforcement learning agent with Reeds-Shepp curves, enabling effective planning across diverse scenarios. HOPE guides the exploration of the reinforcement learning agent by applying an action mask mechanism and employs a transformer to integrate the perceived environmental information with the mask. Our approach achieved higher planning success rates compared with typical rule-based algorithms and traditional reinforcement learning methods, especially in challenging cases.

Examples

Simulation cases

Realworld demo

https://www.youtube.com/watch?v=62w9qhjIuRI

Setup

1. Install conda or miniconda

2. Clone the repo and build the environment

git clone https://github.com/jiamiya/HOPE.git
cd HOPE
conda create -n HOPE python==3.8
conda activate HOPE
pip3 install -r requirements.txt

and install pytorch from https://pytorch.org/.

Usage

Run a pre-trained agent

cd src
python ./evaluation/eval_mix_scene.py ./model/ckpt/HOPE_SAC0.pt --eval_episode 10 --visualize True

You can find some other pre-trained weights in ./src/model/ckpt.

Train the HOPE planner

cd src
python ./train/train_HOPE_sac.py

or

python ./train/train_HOPE_ppo.py

Citation

If you find our work useful, please cite us as

@article{jiang2024hope,
  title={HOPE: A Reinforcement Learning-based Hybrid Policy Path Planner for Diverse Parking Scenarios},
  author={Jiang, Mingyang and Li, Yueyuan and Zhang, Songan and Chen, Siyuan and Wang, Chunxiang and Yang, Ming},
  journal={arXiv preprint arXiv:2405.20579},
  year={2024}
}