Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Generating and Explaining Corner Cases Using Learnt Probabilistic Lane Graphs

Enrik Maci, Rhys Howard, Lars Kunze

TL;DR

This paper introduces Probabilistic Lane Graphs (PLGs) to describe a finite set of lane positions and directions in which vehicles might travel and uses reinforcement learning techniques to modify this policy and to generate realistic and explainable corner case scenarios which can be used for assessing the safety of AV s.

Abstract

Validating the safety of Autonomous Vehicles (AVs) operating in open-ended, dynamic environments is challenging as vehicles will eventually encounter safety-critical situations for which there is not representative training data. By increasing the coverage of different road and traffic conditions and by including corner cases in simulation-based scenario testing, the safety of AVs can be improved. However, the creation of corner case scenarios including multiple agents is non-trivial. Our approach allows engineers to generate novel, realistic corner cases based on historic traffic data and to explain why situations were safety-critical. In this paper, we introduce Probabilistic Lane Graphs (PLGs) to describe a finite set of lane positions and directions in which vehicles might travel. The structure of PLGs is learnt directly from spatio-temporal traffic data. The graph model represents the actions of the drivers in response to a given state in the form of a probabilistic policy. We use reinforcement learning techniques to modify this policy and to generate realistic and explainable corner case scenarios which can be used for assessing the safety of AVs.

Generating and Explaining Corner Cases Using Learnt Probabilistic Lane Graphs

TL;DR

This paper introduces Probabilistic Lane Graphs (PLGs) to describe a finite set of lane positions and directions in which vehicles might travel and uses reinforcement learning techniques to modify this policy and to generate realistic and explainable corner case scenarios which can be used for assessing the safety of AV s.

Abstract

Validating the safety of Autonomous Vehicles (AVs) operating in open-ended, dynamic environments is challenging as vehicles will eventually encounter safety-critical situations for which there is not representative training data. By increasing the coverage of different road and traffic conditions and by including corner cases in simulation-based scenario testing, the safety of AVs can be improved. However, the creation of corner case scenarios including multiple agents is non-trivial. Our approach allows engineers to generate novel, realistic corner cases based on historic traffic data and to explain why situations were safety-critical. In this paper, we introduce Probabilistic Lane Graphs (PLGs) to describe a finite set of lane positions and directions in which vehicles might travel. The structure of PLGs is learnt directly from spatio-temporal traffic data. The graph model represents the actions of the drivers in response to a given state in the form of a probabilistic policy. We use reinforcement learning techniques to modify this policy and to generate realistic and explainable corner case scenarios which can be used for assessing the safety of AVs.
Paper Structure (15 sections, 8 equations, 3 figures, 1 table, 2 algorithms)

This paper contains 15 sections, 8 equations, 3 figures, 1 table, 2 algorithms.

Table of Contents

  1. Introduction
  2. Related Work
  3. Probabilistic Lane Graphs
  4. Definition
  5. PLG Generation
  6. Vehicle Path Planning
  7. Corner Case Generation
  8. Experimental Results
  9. PLG Generation
  10. Corner Case Generation: Quantitative Results
  11. Corner Case Generation: Qualitative Examples
  12. Summary
  13. Conclusion
  14. Discussion & Future Work
  15. Conclusion

Figures (3)

  • Figure 1: A corner case generated using PLG learnt from real-world observations. In this example, the black vehicle (ID 1193) attempted a sharp cut-in in front of another vehicle (ID 1208) leading to a crash event (at $t=39$). Due to the higher velocity of the trailing vehicle a rear-end collision could not be prevented. Note that vehicle 1193 (black) is plotted on top of vehicle 1208.
  • Figure 2: Probabilistic Lane Graph (PLG) for real-world highway data dataset:NGSIM. Left: Entire PLG. Right: Zoomed-in subset of the PLG representing vehicle positions and possible transitions, with edge shading indicating the probability of a vehicle traversing an edge.
  • Figure 3: Corner case scenarios leading to accidents with different numbers of lane changes.