MultiPath: Multiple Probabilistic Anchor Trajectory Hypotheses for Behavior Prediction

Yuning Chai, Benjamin Sapp, Mayank Bansal, Dragomir Anguelov
Proceedings of the Conference on Robot Learning, PMLR 100:86-99, 2020.

Abstract

Predicting human behavior is a difficult and crucial task required for motion planning. It is challenging in large part due to the highly uncertain and multimodal set of possible outcomes in real-world domains such as autonomous driving. Beyond single MAP trajectory prediction [1, 2], obtaining an accurate probability distribution of the future is an area of active interest [3, 4]. We present MultiPath, which leverages a fixed set of future state-sequence anchors that correspond to modes of the trajectory distribution. At inference, our model predicts a discrete distribution over the anchors and, for each anchor, regresses offsets from anchor waypoints along with uncertainties, yielding a Gaussian mixture at each time step. Our model is efficient, requiring only one forward inference pass to obtain multi-modal future distributions, and the output is parametric, allowing compact communication and analytical probabilistic queries. We show on several datasets that our model achieves more accurate predictions, and compared to sampling baselines, does so with an order of magnitude fewer trajectories.

Cite this Paper

BibTeX
@InProceedings{pmlr-v100-chai20a, title = {MultiPath: Multiple Probabilistic Anchor Trajectory Hypotheses for Behavior Prediction}, author = {Chai, Yuning and Sapp, Benjamin and Bansal, Mayank and Anguelov, Dragomir}, booktitle = {Proceedings of the Conference on Robot Learning}, pages = {86--99}, year = {2020}, editor = {Kaelbling, Leslie Pack and Kragic, Danica and Sugiura, Komei}, volume = {100}, series = {Proceedings of Machine Learning Research}, month = {30 Oct--01 Nov}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v100/chai20a/chai20a.pdf}, url = {https://proceedings.mlr.press/v100/chai20a.html}, abstract = {Predicting human behavior is a difficult and crucial task required for motion planning. It is challenging in large part due to the highly uncertain and multimodal set of possible outcomes in real-world domains such as autonomous driving. Beyond single MAP trajectory prediction [1, 2], obtaining an accurate probability distribution of the future is an area of active interest [3, 4]. We present MultiPath, which leverages a fixed set of future state-sequence anchors that correspond to modes of the trajectory distribution. At inference, our model predicts a discrete distribution over the anchors and, for each anchor, regresses offsets from anchor waypoints along with uncertainties, yielding a Gaussian mixture at each time step. Our model is efficient, requiring only one forward inference pass to obtain multi-modal future distributions, and the output is parametric, allowing compact communication and analytical probabilistic queries. We show on several datasets that our model achieves more accurate predictions, and compared to sampling baselines, does so with an order of magnitude fewer trajectories.} }
Endnote
%0 Conference Paper %T MultiPath: Multiple Probabilistic Anchor Trajectory Hypotheses for Behavior Prediction %A Yuning Chai %A Benjamin Sapp %A Mayank Bansal %A Dragomir Anguelov %B Proceedings of the Conference on Robot Learning %C Proceedings of Machine Learning Research %D 2020 %E Leslie Pack Kaelbling %E Danica Kragic %E Komei Sugiura %F pmlr-v100-chai20a %I PMLR %P 86--99 %U https://proceedings.mlr.press/v100/chai20a.html %V 100 %X Predicting human behavior is a difficult and crucial task required for motion planning. It is challenging in large part due to the highly uncertain and multimodal set of possible outcomes in real-world domains such as autonomous driving. Beyond single MAP trajectory prediction [1, 2], obtaining an accurate probability distribution of the future is an area of active interest [3, 4]. We present MultiPath, which leverages a fixed set of future state-sequence anchors that correspond to modes of the trajectory distribution. At inference, our model predicts a discrete distribution over the anchors and, for each anchor, regresses offsets from anchor waypoints along with uncertainties, yielding a Gaussian mixture at each time step. Our model is efficient, requiring only one forward inference pass to obtain multi-modal future distributions, and the output is parametric, allowing compact communication and analytical probabilistic queries. We show on several datasets that our model achieves more accurate predictions, and compared to sampling baselines, does so with an order of magnitude fewer trajectories.
APA
Chai, Y., Sapp, B., Bansal, M. & Anguelov, D.. (2020). MultiPath: Multiple Probabilistic Anchor Trajectory Hypotheses for Behavior Prediction. Proceedings of the Conference on Robot Learning, in Proceedings of Machine Learning Research 100:86-99 Available from https://proceedings.mlr.press/v100/chai20a.html.

Related Material