Application Of Grey Correlation Theory Based On Improved CRITIC Method In Autonomous Vehicles Test And Evaluation

With the rapid development of autonomous driving technology and computer technology,autonomous vehicles have gradually become a research hotspot in recent years.The test and evaluation of autonomous vehicles was critical,because reasonable test scenarios and scientific and objective evaluation methods promote the rapid development of autonomous technology.This paper focused on the quantitative evaluation of autonomous vehicles,and proposed an evaluation method based on the Criteria Importance Though Intercrieria Correlation(CRITIC)grey correlation theory.This paper combines with the National Key Research and Development Program "Research on Construction and Simulation Technology of Hardware in Loop Testing Scenario for Self-Driving Electric Vehicle"(2018YFB0105103).Firstly,the test scenarios of autonomous vehicles were designed and the evaluation index system of autonomous vehicles was established.Secondly,the calculation method of evaluation index weight was determined and the evaluation method for autonomous vehicles was selected.Finally,Pre Scan and Simulink are used for co-simulation to complete the quantitative evaluation of autonomous vehicles.The main contributions of this paper are as follows:The test scenarios of autonomous vehicles divided into two parts: test contents and test environments.The test contents divided the driving behavior of vehicles into basic driving behavior,advanced driving behavior and driving behavior in complex environments;the test environments included road environments,weather and light environments,traffic signal environments,obstacle environments,auditory environments and external interference environments.Based on the existing natural driving data and accident data,dangerous driving scenarios were extracted and analyzed,and several typical test scenarios were listed.The evaluation index system was divided into three layers: the target layer was autonomous vehicles evaluation;the total index layer included driving security,comfortableness,intelligent,and efficiency;and the index layer included 13 detailed indexes.According to the commonly used methods for determining the weight of subjective and objective indicators,a method for calculating the weight of the indicator combining subjective and objective was proposed: the total index layer weight was determined by the Analytic Hierarchy Process(AHP)subjective weighting method,the index layer weight was determined by CRITIC objective weighting method.The gray correlation analysis method was modified which improved the reasonability of the method.In the calculation formula of gray correlation coefficient,the value of resolution coefficient was changed from fixed value to value based on data change.After the combination of typical test scenarios,the simulation tests and analysis evaluation were carried out.The test results showed that the improved grey correlation theory evaluation method had a high consistency with the evaluation results of traditional fuzzy comprehensive evaluation method.The proposed methodology removed the expert scoring part,which reduced the subjectivity.Meanwhile,it saved the cost of inviting a large number of experts to participate in the evaluation and improved the efficiency of test evaluation.