A method for evaluating the complexity of test scenarios for autonomous vehicles

With the continuous development of automotive driving automation, scenario-based automated driving test evaluation methods have become an industry consensus. However, in terms of scenarios, the industry still relies on the subjective experience of experts to formulate evaluation plans, lacking scientific and quantitative evaluation methods, resulting in the problems of limited scenario coverage and low testing efficiency, which affect the mass production process of products. Therefore, this paper proposes a scene complexity evaluation method based on analytic hierarchy process (AHP) and information entropy theory, which realizes the automatic quantitative evaluation of test scene complexity and makes up for the lack of theoretical research on industry test scene. Finally, the method proposed in this paper quantitatively analyzes the complexity of typical scenarios in the Autonomous Driving Evaluation Project of CATARC, summarizes the key factors affecting the complexity of the scenario, and verifies the feasibility and effectiveness of the method.