Test Scenarios Construction Based on Combinatorial Testing Strategy for Automated Vehicles

Abstract

Scenario-based testing is an important verification and certification measure to evaluate the safety of automated vehicles. In view of the existing test scenario composition methods, which may miss some critical scenario problems that have low occurrence probability, we fully combined the ego-vehicle with the possible relative positions and movement directions of surrounding traffic participants based on a complex scenario group. We applied scenario-screening rules to obtain the functional test scenarios with different traffic environments and driving task complexities, which ensured the coverage of the test scenarios and reduced the number of test scenarios. The problem arose that the amount of test cases was too large after the discretized combination of test scenario parameters, so we adopted a three-way combinatorial testing strategy to greatly reduce the number of test cases. Taking the complicated lane changing scenario of the ego-vehicle as an example, the simulation method was adopted, and the critical test cases were obtained by screening through safety indicators. Finally, the K-medoids clustering method was used to further reduce the number of critical test cases, and a pairwise combinatorial test strategy was used to combine dynamic scenario and static scenario elements to obtain critical test cases for closed-road testing.

Keywords

• Automated vehicles
• combinatorial testing strategy
• functional test scenarios
• critical test cases
• safety indicators