Automated vehicle crash sequences: Patterns and potential uses in safety testing

Abstract

With safety being one of the primary motivations for developing automated vehicles (AVs), extensive field and simulation tests are being carried out to ensure AVs can operate safely on roadways. Since 2014, the California Department of Motor Vehicles (DMV) has been collecting AV collision and disengagement reports, which are valuable data sources for studying AV crash patterns. A crash sequence of events describes the AV's interactions with other road users before a collision in a temporal manner. In this study, sequence of events data extracted from California AV collision reports were used to investigate patterns and how they may be used to develop AV test scenarios. Employing sequence analysis methods and clustering, this study evaluated 168 AV crashes (with AV in automatic driving mode before disengagement or collision) reported to the California DMV from 2015 to 2019. Analysis of subsequences showed that the most representative pattern in AV crashes was "collision following AV stop". Analysis of event transition showed that disengagement, as an event in 24% of all studied AV crash sequences, had a transition probability of 68% to an immediate collision. Cluster analysis characterized AV crash sequences into seven groups with distinctive crash dynamic features. Cross-tabulation analysis showed that sequence groups were significantly associated with variables measuring crash outcomes and describing environmental conditions. Crash sequences are useful for developing AV test scenarios. Based on the findings, a scenario-based AV safety testing framework was proposed with sequence of events embedded as a core component.

A Survey on Automated Driving System Testing: Landscapes and Trends

“…It can be seen that there exists only system study for Apollo while bug/issue studies cover the three open-source ADS which we mentioned in §2.3. In addition, more existing works [39][40][41][42][43][44][45][46] perform empirical studies on the disengagement and accident reports for identifying the root causes or failure types, since these investigations are critical to understand the ADS safety performances in the real world.…”

“…The following works [39][40][41][42][43][44][45][46] all perform analysis on the reports from the California Department of Motor Vehicles (CADMV) [37], which involve disengagement and accident records on public roads. Specifically, a disengagement refers to a failure that requires human driver to take over the control of vehicles; an accident refers to a collision with other traffic participants.…”

“…Analysis of accident reports. The following works [43][44][45][46] identify the common accident types from accident reports. In [43], Alambeigi et al utilize probabilistic topic models to identify common crash patterns, e.g., driver-initiated transition crashes, sideswipe crashes during left-side overtakes.…”

“…Boggs et al [44] and Leilabadi et al [45] apply text analysis to the accident reports, and they find that the accidents mostly occur when vehicles run in the automated mode, and the most frequent ADS crash type is the rear-end collision. In [46], Song et al conclude that the most representative accident pattern is the "collision following ADS stop", i.e., an ADS stops suddenly and gets hit by…”

“…The analysis and classification of the disengagements based on different perspectives (e.g., modules) [39][40][41][42] The identification of the common accident types by different methods (e.g., text analysis) [43][44][45][46] other vehicles on the road. Moreover, the probability by which a disengagement can be transited to a collision is 68%.…”