Research On Automatic Test And Evaluation Of Speed Control Strategy For Autonomous Vehicle

Intelligent assisted driving system is more and more widely used in cars,and its speed control determines the safety of autonomous driving vehicles,which has been widely paid attention to.How to use computer system to automatically test and evaluate the speed control strategy of autonomous vehicle,speed up the system maturity and shorten the time to market has become a hot issue in the automotive industry.This paper first analyzes the speed control strategy of autonomous vehicle.ACC and AEB are selected to analyze the control algorithms that affect the speed control and driving safety.Including: ACC system state machine,speed and distance control switching algorithm,start-stop control strategy,and execution control strategy;Dangerous target identification algorithm and control decision algorithm in AEB system.At the same time,the algorithm to deal with the emergency entry of the front vehicle is studied,the testing items in the speed control strategy are refined,and the parameter selection scheme and improvement direction are proposed,which lays a foundation for the automatic testing and optimization of the speed control strategy.Secondly,an automated test hardware-in-the-loop simulation（Hi L）system is developed to meet the requirements of speed control strategy testing and evaluation of autonomous vehicles.System adopts the form of the up and down a machine division,PC with automation scheduling software and test scenario,the deployment of real-time system and evaluation model,complete the automatic invocation of test scenarios,scene interaction parameters and switch,test environment under the machine deployment and testing process control,and test report generation and output,constructed a complete automated test process.Based on the establishment of Hi L test system for autonomous vehicle,test scenarios for speed control strategy were extracted.For ACC speed control algorithm,the function scenes such as cruise control,steady-state following,curve following and cutting out of the vehicle in front are extracted.Considering that the front-vehicle emergency cut-in is the main reason for the dangerous operation of autonomous vehicles,the front-vehicle cut-in function scenarios are analyzed in detail.Based on the natural driving data cut by the vehicle in front,the original scene was de-noised,and then the scene features were extracted,and the driving conditions were reconstructed using k-means clustering method.According to the statistical data of scene elements,the concept of reinforcement coefficient is introduced to obtain the effectiveness of scene design according to the probability of occurrence.In the same way,the test scenario of AEB algorithm is also given.Thirdly,based on the extracted test scenarios and referring to the existing evaluation criteria,a hierarchical test evaluation method from functional scenarios to test scenarios is established.Different weights are given to logical scenes from the level of functional scenes,and weights are given to different test scenes at the level of logical scenes.At the level of test scene,the evaluation indexes are mainly divided into two categories: safety and comfort.Safety indicators include collision time TTC and headway THW.Comfort index includes maximum deceleration,maximum deceleration change rate,relative speed change rate and so on.Based on the above indicators,the evaluation method of autonomous vehicle speed control is formed through the combination of scenarios.Finally,based on the built Hi L test platform and the proposed test scenarios and evaluation methods,the existing speed control strategy algorithm is comprehensively tested,and the algorithm is improved and optimized for the problems in the test.Aiming at the ACC front vehicle entry problem,the slope and velocity following PI parameters of the speed distance control switch line in the dynamics calculation module were selected to calibrate the parameters by using the orthogonal test method.The control strategy can reduce the collision risk and improve the control comfort by predicting the cutting action of the front vehicle in advance.The improved algorithm is tested and compared to verify the usability,accuracy and convenience of the designed automatic test system.