Intelligent Vehicle Environment Perception Based On Visual Radar Information And Control Software Development

With the increase in car ownership in China,Traffic and road safety pressure increasingly become a serious problem.The latest proposed"Made in China 2025" included ten priority areas,and saving and new energy vehicles are among them.Intelligent Network of automotive development goals have been clearly put forward.Intelligent vehicles not only meet the national requirements for the automotive industry,but also good for easing traffic safety.Make intelligent vehicle safety on the roads,the premise is to have an accurate perception and recognition of the traveling environment.Urban roads backdrop,this paper studies intelligent vehicle environment information identification and modeling.At the same time design and develop:ment of intelligent vehicle monitoring and control system software.First,the RANSAC algorithm is simplified,and the least squares fitting lane;A new method is proposed which is based on the"proportion of point on lane line"to distinguish the dotted or solid line.The results show by accurately detecting the lane line,this method has a high recognition rate.Combined with camera calibration results,improved the existing lanes yaw warning model.Using the detection results of the left and right lane line model parameters compute the yaw distance and yaw angle of the vehicle.For the intelligent vehicle path planning and motion control is a necessary condition.Then,the application of the smart mac.hine learning algorithms to detect pedestrians in the vehicle longitudinal direction.Collected a large number of actual road environments pedestrian positive and negative samples,combined with the MIT and INRIA database,uniform sample size and extract positive and negative samples of HOG features and trained cascade classifier and SVM,and the two classifications with the Opencv machine vision library comes with pedestrian detection cascade classifier performance comparison.After the test found that the overall effect of pedestrian detection method HOG feature and SVM better.In addition,the application of monocular vision together with the millimeter-wave radar distance measurement is an obstacle in front of Intelligent Vehicle,by combining visual distance and millimeter-wave radar data,to achieve the Intelligent Vehicle stationary obstacle distance and relative motion detection information,reduce the blind spot detectionFinally,use Visual Studio 2010 development platform and Opencv computer vision library,on the basis of multithreaded programming,completed the overall framework of measurement and control system structures and module function realization.Provide necessary conditions for intelligent vehicle real vehicle test.