Research On Perception Technology Of Parking Space And Path Planning For Automatic Valet Parking System

In recent years,with the increase in car ownership in the world,the problem of parking difficulty has gradually become the main problem which affects people driving.Automatic valet parking is a kind of low-speed local area driverless technology,which can realize the automatic driving and parking functions of driverless vehicles in the parking lot.It is a new technology that effectively solve problem of finding parking space and parking.The perception of the parking space status and the reasonable path planning of driverless vehicles in the parking lot are key technologies of the automatic valet parking system.The research of this thesis based on the perception technology of parking space,the global path planning and the local path planning scheme of driverless vehicles in the automatic valet parking lot.The main contents are as follows:(1)According to the characteristics of the existing experimental parking lot studied in this thesis,the appropriate geomagnetic sensor module and wireless communication module are selected to build the perception system of parking space for the demand of automatic valet parking lot to detect the parking space status and transmit the parking space status information in real-time.The geomagnetic sensor senses the status information of the parking space through the disturbance of the geomagnetic signal caused by the vehicle entering or exiting the parking space,then uploads the information to the background server through the wireless communication module,and uses the designed parking space status monitoring system to display the real-time status information of each parking space in the parking lot.Finally,they will be verified by field vehicle tests.(2)An improved Dijkstra algorithm is proposed for solving the congestion problem of automatic valet parking lot when facing a large traffic flow.The algorithm combines the map information of the automatic valet parking lot and the real-time dynamic influencing factors in the parking lot.It converts the real-time status information of vehicles and parking spaces into basic data for calculation.Finally,this algorithm will be compared with existing path planning algorithms through simulation experiments to verify the effectiveness of this scheme in solving the automatic valet parking lot congestion problem.(3)According to the parking in and out of the parking space,intersection turning and obstacle avoidance conditions during automatic valet parking,different path planning schemes will be designed.The geometric algorithm can be directly used to plan the driving trajectory of parking in and out of the parking space and the intersection turning in advance because the environmental parameters of the parking spaces and intersections are known.Then using the second-order Bezier curve to fit smooth curves for driverless vehicle tracking.Considering the position and geometric parameters of obstacles are unknown in the process of obstacle avoidance,the obstacle avoidance baseline will be planned by the strategy of combining search algorithm and geometric algorithm.Then using the third-order Bezier curve to fit the smooth curve which can be tracked by the driverless vehicle.Finally,the integrated model of the parking lot is used to carry out the simulation of local path planning scheme feasibility.The result shows that the designed local path planning scheme has feasibility and robustness.