Safety risk assessment for autonomous vehicle road testing

Abstract

Objective

Road testing can accelerate the development and validation of autonomous vehicles (AVs). AV road testing can come with high safety risks, particularly in a complex road traffic environment, due to the immaturity of AV technology. A priori safety risk assessments of the road traffic environment before AV road testing are of great importance, allow the quantifying of risk levels in different road scenarios, and provide guidelines for AV road testing in low to high-risk environments.

Methods

This study proposes a framework, namely Safety Risk Assessment for AV road testing (SRAAV), based on the probability and severity of five categories of potential AV accidents. Four groups of influencing factors are considered comprehensively in assessing AV safety risk, and their impacts are quantified using impact coefficients derived from a Bayesian network and empirical AV road testing data. The safety risk is assessed on a road section level, based on which an overall risk level is defined for a corridor and a region. Afterwards, the quantified safety risk is classified into four levels according to expert experience and knowledge, through a questionnaire survey.

Results

Applications of the proposed SRAAV framework are conducted for urban roads in Shanghai, and expressways in Shanghai and Gothenburg. The assessment results are validated using disengagement data from AV road testing. The results show that the SRAAV framework and its models could estimate the safety risk levels of road traffic environments for AV road testing in a sound way and have the flexibility for further extensions to be made.

Conclusions

The framework and assessment results can provide technical support for determining where and when to grant permission for public roads to be used for AV road testing, and how to choose public roads from a low to a high risk level, guaranteeing the safety of AV public road testing.

Keywords:

• Autonomous vehicle road testing
• safety risk assessment
• Bayesian network
• risk levels

Correction Statement

This article has been republished with minor changes. These changes do not impact the academic content of the article.

Additional information

Funding

This research is supported by the National Key R&D Program of China [2019YFE0108300] and the Key Project of Science and Technology Commission of Shanghai Municipality [22dz1203400]. This work reflects only the authors’ views.