Research On The Key Technology Of Automatic Parking System Based On Multi-Source Information Fusion

In recent years,intelligent driving vehicle had become an important development trend in the automotive field,and automatic parking system,as an important part of intelligent driving technology,had also become a hot spot for research.Parking space recognition and path planning are the key technologies of the automatic parking system.The available parking space recognition technology cannot recognize irregular parking spaces,resulting in the waste of parking space resources.In addition,due to the difference of drivers’ parking level,collision and cut-off accidents occur frequently in the parking process.In this paper,parking space recognition and path planning are studied.In the aspect of parking space recognition,in order to solve the problem that the available parking space recognition technology cannot recognize irregular parking spaces,information fusion technology was used to fuse the feature information extracted from millimeter wave radar,camera and wheel speed sensor,to establish the target parking space plane model,and to realize the parking space type recognition by Sugeon-type fuzzy inference with the parking experience of skilled drivers.In terms of path planning,a vehicle kinematic model with the center of the rear axle of the vehicle as the research object was constructed based on the Ackermann steering principle,and considering the obstacle avoidance constraints and safety issues in the parking process,a multi-stage parking path was planned with the goal of collision-free and parking into the parking space under parallel and perpendicular parking conditions,and the parking path planning method was simulated on the MATLAB platform.Considering the problem of discontinuous curvature of the parking path based on circular straight lines,a Bezier curve was used to smooth it,a new gridbased representation(2n×2n grid)was used to describe the parking environment,the wheel steering performance constraints and obstacle avoidance constraints were combined,the length of the collision-free Bezier curve parking path determined by the control points was used as the optimization criterion,and an improved genetic algorithm was used to the control points of the Bessel curve were optimized,and a parking path with continuous curvature and satisfying the constraints was planned by using the nature of the Bessel curve.The simulation was performed on MATLAB,and the results show that the path optimization method based on the improved genetic algorithm can achieve the optimization of the circular linear path.