Generating Scenarios with Diverse Pedestrian Behaviors for Autonomous Vehicle Testing
(2022) Conference on Robot Learning In Proceedings of Machine Learning Research 164. p.1247-1258- Abstract
- There exist several datasets for developing self-driving car methodologies. Manually collected datasets impose inherent limitations on the variability of test cases and it is particularly difficult to acquire challenging scenarios, e.g. ones involving collisions with pedestrians. A way to alleviate this is to consider automatic generation of safety-critical scenarios for autonomous vehicle (AV) testing. Existing approaches for scenario generation use heuristic pedestrian behavior models. We instead propose a framework that can use state-of-the-art pedestrian motion models, which is achieved by reformulating the problem as learning where to place pedestrians such that the induced scenarios are collision prone for a given AV. Our pedestrian... (More)
- There exist several datasets for developing self-driving car methodologies. Manually collected datasets impose inherent limitations on the variability of test cases and it is particularly difficult to acquire challenging scenarios, e.g. ones involving collisions with pedestrians. A way to alleviate this is to consider automatic generation of safety-critical scenarios for autonomous vehicle (AV) testing. Existing approaches for scenario generation use heuristic pedestrian behavior models. We instead propose a framework that can use state-of-the-art pedestrian motion models, which is achieved by reformulating the problem as learning where to place pedestrians such that the induced scenarios are collision prone for a given AV. Our pedestrian initial location model can be used in conjunction with any goal driven pedestrian model which makes it possible to challenge an AV with a wide range of pedestrian behaviors – this ensures that the AV can avoid collisions with any pedestrian it encounters. We show that it is possible to learn a collision seeking scenario generation model when both the pedestrian and AV are collision avoiding. The initial location model is conditioned on scene semantics and occlusions to ensure semantic and visual plausibility, which increases the realism of generated scenarios. Our model can be used to test any AV model given sufficient constraints. (Less)
Please use this url to cite or link to this publication:
https://lup.lub.lu.se/record/3590c5e6-8b7e-4c39-9025-f9df7f747fd6
- author
- Priisalu, Maria LU ; Pirinen, Aleksis LU ; Paduraru, Ciprian and Sminchisescu, Cristian LU
- organization
- publishing date
- 2022
- type
- Chapter in Book/Report/Conference proceeding
- publication status
- published
- subject
- keywords
- reinforcement learning, autonomous vehicles, scenario generation, testing autonomous vehicles
- host publication
- Conference on Robot Learning, 8-11 November 2021, London, UK
- series title
- Proceedings of Machine Learning Research
- volume
- 164
- pages
- 12 pages
- publisher
- ML Research Press
- conference name
- Conference on Robot Learning
- conference location
- London, United Kingdom
- conference dates
- 2021-11-08
- external identifiers
-
- scopus:85171741155
- ISSN
- 2640-3498
- project
- Modelling Pedestrians in Autonomous Vehicle Testing
- Semantic Mapping and Visual Navigation for Smart Robots
- language
- English
- LU publication?
- yes
- id
- 3590c5e6-8b7e-4c39-9025-f9df7f747fd6
- date added to LUP
- 2022-09-21 16:39:05
- date last changed
- 2023-12-28 12:39:07
@inproceedings{3590c5e6-8b7e-4c39-9025-f9df7f747fd6,
abstract = {{There exist several datasets for developing self-driving car methodologies. Manually collected datasets impose inherent limitations on the variability of test cases and it is particularly difficult to acquire challenging scenarios, e.g. ones involving collisions with pedestrians. A way to alleviate this is to consider automatic generation of safety-critical scenarios for autonomous vehicle (AV) testing. Existing approaches for scenario generation use heuristic pedestrian behavior models. We instead propose a framework that can use state-of-the-art pedestrian motion models, which is achieved by reformulating the problem as learning where to place pedestrians such that the induced scenarios are collision prone for a given AV. Our pedestrian initial location model can be used in conjunction with any goal driven pedestrian model which makes it possible to challenge an AV with a wide range of pedestrian behaviors – this ensures that the AV can avoid collisions with any pedestrian it encounters. We show that it is possible to learn a collision seeking scenario generation model when both the pedestrian and AV are collision avoiding. The initial location model is conditioned on scene semantics and occlusions to ensure semantic and visual plausibility, which increases the realism of generated scenarios. Our model can be used to test any AV model given sufficient constraints.}},
author = {{Priisalu, Maria and Pirinen, Aleksis and Paduraru, Ciprian and Sminchisescu, Cristian}},
booktitle = {{Conference on Robot Learning, 8-11 November 2021, London, UK}},
issn = {{2640-3498}},
keywords = {{reinforcement learning; autonomous vehicles; scenario generation; testing autonomous vehicles}},
language = {{eng}},
pages = {{1247--1258}},
publisher = {{ML Research Press}},
series = {{Proceedings of Machine Learning Research}},
title = {{Generating Scenarios with Diverse Pedestrian Behaviors for Autonomous Vehicle Testing}},
volume = {{164}},
year = {{2022}},
}