Research On The Personalized Lane-Change Assistance System With Driving Style Identification

Lane-Change Assistance(LCA)system is a typical lateral intelligent drving assistance system with functions of lane-change warning,trajectory planning,and trajectory tracking,which is beneficial to reduce accidents,improve driving safety,alleviate driving fatigue,and improve road capacity.However the current LCA system neglects the influence and interference of driving style,and the single designed mode can not meet the individual differences of drivers,which reduces the acceptance and satisfaction of drivers to LCA system.In the closed-loop system of human-vehicle-environment,the driver is the weakest part with strong complexity and uncertainty.Therefore we need to deeply study the driving style and establish its representation scheme.Considering the driving style in the design of decisionmaking mechanism and control system is an important research direction of LCA system in the future,and it is also the premise and foundation for large-scale promotion and application of LCA system.This paper carries out a study on the personalized lane-change assistance system with driving style identification supported by the National Key R&D Program of China(2016YFB0100904),National Natural Science Foundation of China(51775235),and Jilin Province Science and Technology Development Plan Projects(20170101138JC).First,the driver's driving behavior data are collected based on the real vehicle driving data acquisition platform.The identification of driving style is realized using a back-propagation(BP)neural network optimized by a particle swarm optimization(PSO)algorithm.Next,the feasibility of a lane change based on surrounding multi-vehicle information is analyzed comprehensively from the perspective of collision-avoidance and vehicle-following.And then a simple sinusoidal lane-change model is adopted to realize personalized lane-change trajectory planning,and the trajectory tracking controller is designed by using the preview optimal curvature model and the model predictive control(MPC)algorithm.Finally,the correctness and superiority of the vehicle lane-change assistance system based on driving style is verified by conducting computer simulations and real vehicle test under typical conditions.The main research contents of this paper include the following four aspects:(1)Data acquisition and analysis of driver behaviorFirst,a real vehicle driving data acquisition platform is built based on RT3000&RTRange high-precision integrated navigation system,ESR millimeter wave radar,camera,Micro Auto Box,RTMaps,d SPACE Control Desk and so on.87 drivers are recruited in the society and driver's driving behavior data are collected under typical working conditions.Then,the factor analysis is used to simplify the data dimensions,and the fuzzy C-means(FCM)clustering algorithm is adopted to classify the driving style into cautious,general and aggressive.Finally the identification of driving style is realized using a BP neural network optimized by a PSO algorithm.(2)Personalized lane-change warning strategy for intelligent vehiclesThe lane-change risks are comprehensively analyzed between the host vehicle and surrounding vehicles,and a personalized lane-change warning scheme is designed based on driving style.On the premise of meeting the requirements of obstacle-avoidance and vfollowing,a personalized minimum safe distance model for lane changing is established considering the influence of driver's driving style on lane-change distance.It can comprehensively evaluate lane-change safety and improve the adaptability of lane-change warning system to different drivers.A personalized lane-change warning strategy is anlayzed and verified based on the Matlab/Simulink-Car Sim joint simulation platform.(3)Personalized lane-change trajectory planning and tracking for intelligent vehiclesBased on the lane-change characteristics of drivers with different driving style,personalized lane-change trajectory planning strategy is designed to improve the acceptance of lane-changing assistance system by different drivers under two scenarios of straight road and curve road.Trajectory tracking controller is designed based on preview optimal curvature model and MPC theory to realize autonomous lane changing.Personalized lane-change trajectory planning and tracking is anlayzed and verified based on the Matlab/SimulinkCar Sim joint simulation platform.(4)Verification and analysis of real vehicle testA multi-source information perception system for intelligent vehicles is established based on real vehicle veri fication platform,and some reasonable working conditions are designed to verify the identification model of driving style,personalized lane-change warning strategy,personalized lane-changing trajectory planning and tracking.The results prove the effectiveness,accuracy,and feasibility of personalized lane-change assistance system with driving style identification.