Research On Multi-sensor Reversing Environment Perception And Speed Control For Electric Vehicle

With the rising of the prices of oil caused by the depletion of oil resources in the world,as well as the emissions of the greenhouse gas,new energy vehicles are more and more noticed by people.Especially,electric vehicles have been favored by many consumers for their advantages of no pollution,low noise and energy saving,and have gradually become the mainstream trend of future vehicle development.With the development of electric vehicles technology,the increasing of the endurance mileage and the support of policy,the number of electric vehicles is increasing year by year.However,more and more car accidents were cuased,due to the increasing number of cars and drivers and the complexity of the environment of the vehicle.As automotive electronic technology and information processing technology have been developed.intelligent auxiliary reversing equipment provides a better solution to these pronlems.Therefore,this paper focused on the intelligent vehicle reversing and speed control of electric vehicle,and presented a new method of active safety control for vehicle reversing based on multi-heterogeneous environment.By using a variety of different types of sensors to obtain the environmental information,multi-heterogeneous reversing environment perception of electric vehicle was achieved.The main works can be described as follows:Firstly,we built a multi-sensor environment perception platform that included binocular cameras,ultrasonic rangefinders and laser rangefinders.The binocular cameras were firstly used to get the vision information of a reversing vehicle,which was helpful to expand the visual field and to reduce the blind area and reduce the driver’s burden.Ultrasonic range finders and laser range finders were mainly used to measure the distance information between vehicle and obstacles.Then,the vision information collected by binocular cameras was processed based on the principle of binocular vision.In this step,the intrinsic,extrinsic parameters and distortion parameters of binocular cameras were obtained by using the basic principle of binocular vision.By busing these parameters,binocular images were corrected and the disparity of the target was computed.At last,the distance between the vehicle and obstacles was obtained by using triangulation ranging.Meanwhile,the target with the greatest threat was detected and its position was calculated.Secondly,according to the distance information obtained by the binocular vision,ultrasonic range finders and laser range finders,we used kalman filter to remove the noise contained in the information.Then,federated filtering framework of information fusion was introduced to reduce the error and improve the reliability of our system.Thirdly,we proposed a novel method that combined low rank representation with unscented particle filter for target tracking and recognition whose initial position was calculated by binocular detection.In this paper,we conducted experiments to verify the tracking and recognition results by using the data of the standard library and the data collected in the actual reversing environment.In addition,a comparison and analysis between the proposed algorithm and other algorithms were made.The experiment results shown that the proposed algorithm in this paper was improved by more than 3 times.Finally,we used the fuzzy control theory to establish the speed control model of electric vehicle.According to the results of information fusion and target recognition and tracking,the simulation of electric vehicle speed control was carried out.Based on the analysis of the different speed control curves,we can get the conclusion that the speed control method of electric vehicle based on multi-heterogeneous environment was feasible.