Fast Risk Assessment for Autonomous Vehicles Using Learned Models of Agent Futures
This paper presents fast non-sampling based methods to assess the risk of trajectories for autonomous vehicles when probabilistic predictions of other agents' futures are generated by deep neural networks (DNNs). The presented methods address a wide range of representations for uncertain pre...
| Main Authors: | , , , |
|---|---|
| Other Authors: | |
| Format: | Article |
| Language: | English |
| Published: |
Robotics: Science and Systems Foundation
2022
|
| Online Access: | https://hdl.handle.net/1721.1/145543 |
| _version_ | 1826213925112250368 |
|---|---|
| author | Wang, Allen Huang, Xin Jasour, Ashkan Williams, Brian |
| author2 | Massachusetts Institute of Technology. Department of Aeronautics and Astronautics |
| author_facet | Massachusetts Institute of Technology. Department of Aeronautics and Astronautics Wang, Allen Huang, Xin Jasour, Ashkan Williams, Brian |
| author_sort | Wang, Allen |
| collection | MIT |
| description | This paper presents fast non-sampling based methods to assess the risk of
trajectories for autonomous vehicles when probabilistic predictions of other
agents' futures are generated by deep neural networks (DNNs). The presented
methods address a wide range of representations for uncertain predictions
including both Gaussian and non-Gaussian mixture models for predictions of both
agent positions and controls. We show that the problem of risk assessment when
Gaussian mixture models (GMMs) of agent positions are learned can be solved
rapidly to arbitrary levels of accuracy with existing numerical methods. To
address the problem of risk assessment for non-Gaussian mixture models of agent
position, we propose finding upper bounds on risk using Chebyshev's Inequality
and sums-of-squares (SOS) programming; they are both of interest as the former
is much faster while the latter can be arbitrarily tight. These approaches only
require statistical moments of agent positions to determine upper bounds on
risk. To perform risk assessment when models are learned for agent controls as
opposed to positions, we develop TreeRing, an algorithm analogous to tree
search over the ring of polynomials that can be used to exactly propagate
moments of control distributions into position distributions through nonlinear
dynamics. The presented methods are demonstrated on realistic predictions from
DNNs trained on the Argoverse and CARLA datasets and are shown to be effective
for rapidly assessing the probability of low probability events. |
| first_indexed | 2024-09-23T15:57:01Z |
| format | Article |
| id | mit-1721.1/145543 |
| institution | Massachusetts Institute of Technology |
| language | English |
| last_indexed | 2024-09-23T15:57:01Z |
| publishDate | 2022 |
| publisher | Robotics: Science and Systems Foundation |
| record_format | dspace |
| spelling | mit-1721.1/1455432022-10-02T05:18:52Z Fast Risk Assessment for Autonomous Vehicles Using Learned Models of Agent Futures Wang, Allen Huang, Xin Jasour, Ashkan Williams, Brian Massachusetts Institute of Technology. Department of Aeronautics and Astronautics This paper presents fast non-sampling based methods to assess the risk of trajectories for autonomous vehicles when probabilistic predictions of other agents' futures are generated by deep neural networks (DNNs). The presented methods address a wide range of representations for uncertain predictions including both Gaussian and non-Gaussian mixture models for predictions of both agent positions and controls. We show that the problem of risk assessment when Gaussian mixture models (GMMs) of agent positions are learned can be solved rapidly to arbitrary levels of accuracy with existing numerical methods. To address the problem of risk assessment for non-Gaussian mixture models of agent position, we propose finding upper bounds on risk using Chebyshev's Inequality and sums-of-squares (SOS) programming; they are both of interest as the former is much faster while the latter can be arbitrarily tight. These approaches only require statistical moments of agent positions to determine upper bounds on risk. To perform risk assessment when models are learned for agent controls as opposed to positions, we develop TreeRing, an algorithm analogous to tree search over the ring of polynomials that can be used to exactly propagate moments of control distributions into position distributions through nonlinear dynamics. The presented methods are demonstrated on realistic predictions from DNNs trained on the Argoverse and CARLA datasets and are shown to be effective for rapidly assessing the probability of low probability events. 2022-09-21T15:42:32Z 2022-09-21T15:42:32Z 2020 2022-09-21T15:33:11Z Article http://purl.org/eprint/type/ConferencePaper https://hdl.handle.net/1721.1/145543 Wang, Allen, Huang, Xin, Jasour, Ashkan and Williams, Brian. 2020. "Fast Risk Assessment for Autonomous Vehicles Using Learned Models of Agent Futures." Robotics: Science and Systems XVI. en 10.15607/RSS.2020.XVI.089 Robotics: Science and Systems XVI Creative Commons Attribution-Noncommercial-Share Alike http://creativecommons.org/licenses/by-nc-sa/4.0/ application/pdf Robotics: Science and Systems Foundation arXiv |
| spellingShingle | Wang, Allen Huang, Xin Jasour, Ashkan Williams, Brian Fast Risk Assessment for Autonomous Vehicles Using Learned Models of Agent Futures |
| title | Fast Risk Assessment for Autonomous Vehicles Using Learned Models of Agent Futures |
| title_full | Fast Risk Assessment for Autonomous Vehicles Using Learned Models of Agent Futures |
| title_fullStr | Fast Risk Assessment for Autonomous Vehicles Using Learned Models of Agent Futures |
| title_full_unstemmed | Fast Risk Assessment for Autonomous Vehicles Using Learned Models of Agent Futures |
| title_short | Fast Risk Assessment for Autonomous Vehicles Using Learned Models of Agent Futures |
| title_sort | fast risk assessment for autonomous vehicles using learned models of agent futures |
| url | https://hdl.handle.net/1721.1/145543 |
| work_keys_str_mv | AT wangallen fastriskassessmentforautonomousvehiclesusinglearnedmodelsofagentfutures AT huangxin fastriskassessmentforautonomousvehiclesusinglearnedmodelsofagentfutures AT jasourashkan fastriskassessmentforautonomousvehiclesusinglearnedmodelsofagentfutures AT williamsbrian fastriskassessmentforautonomousvehiclesusinglearnedmodelsofagentfutures |