Indoor Parking Method Based On Cooperation Of Intelligent Vehicle And Parking Lot

With the increasing of vehicles,more and more large indoor parking lots have been built,this situation makes the parking cruise problem no longer only occur outside the parking lot.It also takes a lot of time and effort to find and park in the parking lot.The existing solutions generally only have the function of parking guidance,and the degree of information and intelligence is not enough.Therefore,using the internet of vehicle and indoor positioning technologies to enhance the informatization and intelligence degree of parking system has important research significance and practical value to effectively solve the problem of parking cruise,improve parking efficiency,and achieve the purpose of reducing resource consumption and traffic pressure.In order to solve the above problems,this thesis designs an indoor parking system based on cooperation of intelligent vehicle and parking lot.The system is composed of parking lot sensor group,parking lot server and intelligent vehicle terminal.The communication between the intelligent vehicle terminal and the parking lot is established by the internet of vehicle technology,and the sensor and the parking lot server are deployed in the parking lot to complete the indoor vehicle positioning and parking space management.The intelligent vehicle terminal initiates a parking/leaving request and a positioning request to the parking lot,and the parking lot server provides the parking lot real-time map,parking space information and positioning information to the intelligent vehicle terminal,so as to complete the parking navigation or automatic parking in the indoor parking scene through the cooperation of the intelligent vehicle terminal and the parking lot.Indoor vehicle target detection and location is the most critical technology in the above system.In this thesis,the indoor vehicle detection and location method based on multi-vision sensors is studied.the static background and gaussian mixture background model are used for the local scene with different light conditions.It can better deal with the scene with complex light conditions such as the entrance and exit of the parking lot,use the principle of chromaticity invariance to deal with the vehicle shadow,use the volume constraint to deal with the visual projection error,and effectively improve the positioning accuracy.Using color and texture features to assist data association is more reliable than the traditional data association method based on probability statistics,and it can also accurately associate target measurement and trajectory in multi-target scene.Kalman filter is used to estimate the target position optimally under the global map to improve the accuracy and stability of positioning and tracking.In this thesis,the effectiveness of the proposed method is verified by simulation experiments,and the collaborative parking system software is developed according to the above architecture.The effectiveness of the proposed method and the developed system is also verified by the miniaturized intelligent vehicle platform and the real parking lot scene experiment.