Intersection Two-Vehicle Crash Scenario Specification for Automated Vehicle Safety Evaluation Using Sequence Analysis and Bayesian Networks

TL;DR

This paper proposes a method for specifying intersection two-vehicle crash scenarios using sequence analysis and Bayesian networks, enabling detailed safety evaluation based on real crash data and operational conditions.

Contribution

It introduces a novel approach combining crash sequence encoding with Bayesian network modeling to characterize and specify crash scenarios for automated vehicle safety testing.

Findings

Identified 55 crash sequence types from real-world data.

Developed a Bayesian network model linking crash sequences with outcomes and conditions.

Enabled scenario specification through probabilistic querying of the model.

Abstract

This paper develops a test scenario specification procedure using crash sequence analysis and Bayesian network modeling. Intersection two-vehicle crash data was obtained from the 2016 to 2018 National Highway Traffic Safety Administration Crash Report Sampling System database. Vehicles involved in the crashes are specifically renumbered based on their initial positions and trajectories. Crash sequences are encoded to include detailed pre-crash events and concise collision events. Based on sequence patterns, the crashes are characterized as 55 types. A Bayesian network model is developed to depict the interrelationships among crash sequence types, crash outcomes, human factors, and environmental conditions. Scenarios are specified by querying the Bayesian network conditional probability tables. Distributions of operational design domain attributes - such as driver behavior, weather, lighting condition, intersection geometry, traffic control device - are specified based on conditions of sequence types. Also, distribution of sequence types is specified on specific crash outcomes or combinations of operational design domain attributes.