Safety risk assessment for autonomous vehicle road testing

Area of Science:

• Autonomous Systems Engineering
• Traffic Safety Analysis
• Risk Management

Background:

• Autonomous vehicle (AV) road testing is crucial for development and validation.
• Immature AV technology presents significant safety risks in complex traffic environments.
• Proactive safety risk assessment is essential for guiding AV road testing.

Purpose of the Study:

• To introduce a novel framework, Safety Risk Assessment for AV road testing (SRAAV), for evaluating road traffic environments.
• To quantify safety risks associated with potential AV accidents and influencing factors.
• To provide a structured approach for risk-based AV road testing deployment.

Main Methods:

• Developed the SRAAV framework based on accident probability and severity.
• Utilized a Bayesian network and empirical data to quantify risk factors.
• Assessed safety risks at road section, corridor, and regional levels.
• Classified quantified risks into four levels using expert-driven surveys.

Main Results:

• Applied the SRAAV framework to urban and expressway scenarios in Shanghai and Gothenburg.
• Validated assessment results using real-world AV road testing disengagement data.
• Demonstrated the framework's capability to accurately estimate safety risk levels.

Conclusions:

• The SRAAV framework provides a sound method for estimating road traffic safety risks for AV testing.
• Results support informed decisions on granting permissions for AV road testing on public roads.
• The framework offers flexibility for future enhancements and broader applications.