Safe Local Motion Planning with Self-Supervised Freespace Forecasting

By Peiyun Hu, Aaron Huang, John Dolan, David Held, and Deva Ramanan

Citing us

You can find our paper on CVF Open Access. If you find our work useful, please consider citing:

@inproceedings{hu2021safe,
  title={Safe Local Motion Planning with Self-Supervised Freespace Forecasting},
  author={Hu, Peiyun and Huang, Aaron and Dolan, John and Held, David and Ramanan, Deva},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  pages={12732--12741},
  year={2021}
}

Setup

• Download nuScenes dataset, including the CANBus extension, as we will use the recorded vehicle state data for trajectory sampling. (Tip: the code assumes they are stored under /data/nuscenes.)
• Install packages and libraries (via conda if possible), including torch, torchvision, tensorboard, cudatoolkit-11.1, pcl>=1.9, pybind11, eigen3, cmake>=3.10, scikit-image, nuscenes-devkit. (Tip: verify location of python binary with which python.)
• Compile code for Lidar point cloud ground segmentation under lib/grndseg using CMake.

Preprocessing

• Run preprocess.py to generate ground segmentations
• Run precast.py to generate future visible freespace maps
• Run rasterize.py to generate BEV object occupancy maps and object "shadow" maps.

Training

Refer to train.py.

Testing

Refer to test.py.

Acknowledgements

Thanks @tarashakhurana for help with README.