Research On Intelligent Vehicle Obstacle Avoidance Planning System Based On Multi-Sensor Fusion

The advancement of intelligent driving technology not only replaces traditional vehicle driving style,but also relies on advanced computer technology to improve the safety and energy efficiency of moving vehicles and enhance logistics efficiency.However,current intelligent vehicles are human-based assisted driving,with some potential hazards in driving,such as lack of reaction to emergency situations,absence of corresponding regulatory schemes and periodic system breakdowns,etc.Meanwhile,the simple use of a single in-vehicle sensor is not sufficient to tackle more overall issues.This study proposed a multi-sensor fusion-based key technology for the active obstacle avoidance of intelligent vehicles,which improved the safety of humans,coupled with the precise sensing information and the scientific control algorithms and made the vehicle driving more stable.In this study,the multi-sensor fusion technology,obstacle avoidance planning algorithm and control algorithm were thoroughly researched and designed,and combined with application requirements of the Intelligent Automotive Obstacle Avoidance Planning System(IAOAPS),the multi-sensor fusion technology was implemented in it and hardware architecture was built with the high-precision combined navigation INS-570 D,77GHz MMW radar,solid state Li DAR RS-Li DAR-M1,MIC-770V2 IPC and V5-7 smart vehicle.Software framework of environment sensing subsystem,real-time positioning subsystem,decision-making planning subsystem and Linear Control subsystem was designed,the functionality and reliability of the system was verified,and the multi-sensor fusion based IAOAPS was realized.At the outset,the study background,the study significances,the current state of national and international research,current situation of the multi-sensors fusion research and obstacle avoidance planning algorithms of intelligent vehicles with multi-sensor data fusion were clearly presented and illustrated,and also analysed the benefits of technical development of obstacle avoidance planning based on the fusion of multi-sensors data.Secondly,the operating theory and mathematical model of the multi-sensors are used in the intelligent vehicle,the conversion of the coordinate system between the sensors and the local vehicle,the principle of multi-sensor data fusion technology and the time and space synchronic algorithm are deeply analysed and studied.The focus was on Lidar’s 3D Point Cloud processing,Radar’s filtering of the target objects,distance and velocity capture,the data fusion algorithms on the basis of Kalman filtering(KF)and extended Kalman filtering(EKF).Then,the smart vehicle avoidance planning algorithm,path tracking algorithm and control execution algorithm are further studied and analysed.The focus is on GPS-RTK to achieve the real-time localisation and time synchronisation,and the establishment of a quintic polynomial function with time as independent variable function-based obstacle avoidance path based on the critical safety distance established by multi-sensor post-fusion data.Decoupling intelligent vehicle the lateral and longitudinal control for better control based on the vehicle Dynamics model and frenet coordinate system,and the datas of lateral LQR control and longitudinal dual PID control were interacted with its execution structure by using Scoket CAN separately.Finally,to further demonstrate the theory and methodology of this study,software and hardware platforms for simulation testing and actual vehicle trials were built by combining previous multi-sensor fusion technology and obstacle avoidance planning algorithms,and the simulated and real vehicle trials are conducted to obtain the effect of this vehicle controlled by different algorithms,as well as the corresponding analysis and summary of the trial results and trial errors.