Learning to Predict Vehicle Trajectories with Model-based Planning

Haoran Song, Di Luan, Wenchao Ding, Michael Y Wang, Qifeng Chen
Proceedings of the 5th Conference on Robot Learning, PMLR 164:1035-1045, 2022.

Abstract

Predicting the future trajectories of on-road vehicles is critical for autonomous driving. In this paper, we introduce a novel prediction framework called PRIME, which stands for Prediction with Model-based Planning. Unlike recent prediction works that utilize neural networks to model scene context and produce unconstrained trajectories, PRIME is designed to generate accurate and feasibility-guaranteed future trajectory predictions. PRIME guarantees the trajectory feasibility by exploiting a model-based generator to produce future trajectories under explicit constraints and enables accurate multimodal prediction by utilizing a learning-based evaluator to select future trajectories. We conduct experiments on the large-scale Argoverse Motion Forecasting Benchmark, where PRIME outperforms the state-of-the-art methods in prediction accuracy, feasibility, and robustness under imperfect tracking.

Cite this Paper

BibTeX
@InProceedings{pmlr-v164-song22a, title = {Learning to Predict Vehicle Trajectories with Model-based Planning}, author = {Song, Haoran and Luan, Di and Ding, Wenchao and Wang, Michael Y and Chen, Qifeng}, booktitle = {Proceedings of the 5th Conference on Robot Learning}, pages = {1035--1045}, year = {2022}, editor = {Faust, Aleksandra and Hsu, David and Neumann, Gerhard}, volume = {164}, series = {Proceedings of Machine Learning Research}, month = {08--11 Nov}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v164/song22a/song22a.pdf}, url = {https://proceedings.mlr.press/v164/song22a.html}, abstract = {Predicting the future trajectories of on-road vehicles is critical for autonomous driving. In this paper, we introduce a novel prediction framework called PRIME, which stands for Prediction with Model-based Planning. Unlike recent prediction works that utilize neural networks to model scene context and produce unconstrained trajectories, PRIME is designed to generate accurate and feasibility-guaranteed future trajectory predictions. PRIME guarantees the trajectory feasibility by exploiting a model-based generator to produce future trajectories under explicit constraints and enables accurate multimodal prediction by utilizing a learning-based evaluator to select future trajectories. We conduct experiments on the large-scale Argoverse Motion Forecasting Benchmark, where PRIME outperforms the state-of-the-art methods in prediction accuracy, feasibility, and robustness under imperfect tracking.} }
Endnote
%0 Conference Paper %T Learning to Predict Vehicle Trajectories with Model-based Planning %A Haoran Song %A Di Luan %A Wenchao Ding %A Michael Y Wang %A Qifeng Chen %B Proceedings of the 5th Conference on Robot Learning %C Proceedings of Machine Learning Research %D 2022 %E Aleksandra Faust %E David Hsu %E Gerhard Neumann %F pmlr-v164-song22a %I PMLR %P 1035--1045 %U https://proceedings.mlr.press/v164/song22a.html %V 164 %X Predicting the future trajectories of on-road vehicles is critical for autonomous driving. In this paper, we introduce a novel prediction framework called PRIME, which stands for Prediction with Model-based Planning. Unlike recent prediction works that utilize neural networks to model scene context and produce unconstrained trajectories, PRIME is designed to generate accurate and feasibility-guaranteed future trajectory predictions. PRIME guarantees the trajectory feasibility by exploiting a model-based generator to produce future trajectories under explicit constraints and enables accurate multimodal prediction by utilizing a learning-based evaluator to select future trajectories. We conduct experiments on the large-scale Argoverse Motion Forecasting Benchmark, where PRIME outperforms the state-of-the-art methods in prediction accuracy, feasibility, and robustness under imperfect tracking.
APA
Song, H., Luan, D., Ding, W., Wang, M.Y. & Chen, Q.. (2022). Learning to Predict Vehicle Trajectories with Model-based Planning. Proceedings of the 5th Conference on Robot Learning, in Proceedings of Machine Learning Research 164:1035-1045 Available from https://proceedings.mlr.press/v164/song22a.html.

Related Material