An Improved Grey Relational Theory Evaluation Method: Considering the Comprehensive Performance of Autonomous Vehicles in Virtual Test

Abstract

Reasonable test scenarios and objective evaluation methods can rapidly promote the development of autonomous vehicle technology. A new quantitative evaluation method for the comprehensive performance of autonomous vehicle is proposed in this paper. First, different test environments and test contents are combined to obtain vehicle test scenarios of different complexity. Then, the evaluation index system of autonomous vehicle is divided into target layer, total index layer, and index layer. After that, the weights of the index layer are determined by the objective weight method of Criteria Importance though Intercriteria Correlation (CRITIC) method, and the total weights of index layer are determined by the analytic hierarchy process (AHP) of subjective weight method. Finally, the improved grey relational theory method is used to quantitatively evaluate autonomous vehicles from four aspects: driving safety, riding comfort, intelligence, and efficiency. The quantitative evaluation of autonomous vehicles can reduce the influence of abnormal data on the correlation degree and increase the robustness of the evaluation algorithm. The evaluation results of the proposed method and the traditional fuzzy comprehensive evaluation method are compared by simulation experiment and evaluation. The results show that the proposed evaluation method in this paper is more objective and reasonable, which can quantitatively evaluate the comprehensive performance of autonomous vehicles.