End-to-end Birds-eye-view Flow Estimation for Autonomous Driving

Links: arxiv, bibtex

Abtract

In autonomous driving, accurately estimating the state of surrounding obstacles is critical for safe and robust path planning. However, this perception task is difficult, particularly for generic obstacles/objects, due to appearance and occlusion changes. To tackle this problem, we propose an end-to-end deep learning framework for LIDAR-based flow estimation in bird’s eye view (BeV). Our method takes consecutive point cloud pairs as input and produces a 2-D BeV flow grid describing the dynamic state of each cell. The experimental results show that the proposed method not only estimates 2-D BeV flow accurately but also improves tracking performance of both dynamic and static objects.

Bibtex

@inproceedings{lee2020endtoend,
    title={End-to-end Birds-eye-view Flow Estimation for Autonomous Driving},
    author={Kuan-Hui Lee and Matthew Kliemann and Adrien Gaidon and Jie Li and Chao Fang and Sudeep Pillai and Wolfram Burgard},
    booktitle={IROS},
    year={2020},
}