Research On Lateral Obstacle Avoidance Strategy Of Intelligent Vehicles Based On Vision And Lidar

With the rapid development of society and economy,automobile has became an indispensable part of people's lives.But the problems caused by the surge in car ownership has gradually become the focus of attention,such as energy depletion,environmental pollution and traffic congestion.Which is the most serious problem of traffic accidents,intelligent technology(automatic driving and advanced driving assistance systems)by domestic scholars as the most promising approach to solve this problems.Intelligent driverless system is mainly composed of environmental awareness,task decision,planning control and bottom execution.The research of intelligent vehicle lateral obstacle avoidance control based on vision and lidar is mainly from three aspects.(1)the target detection research of the obstacle vehicle is carried out by the monocular industrial camera as the main sensor.First compare the current target detection algorithm in order to show the advantages of the AdaBoost algorithm,and then elaborates the basic principle of Haar-like features and AdaBoost algorithm for training the classifier.Next introduce the whole classifier training process and obtain the classifier model file.Finally through the simulation to verify the accuracy of the algorithm for vehicle detection in actual road conditions.(2)the method of vehicle detection for visual and lidar information fusion is studied.First introduces the process of information fusion algorithm and image preprocessing,and then the fusion method of space and time in the process of multi-sensor information fusion is elaborated.After that the specific implementation method of vehicle detection combined with the acquisition of position information of lidar and AdaBoost algorithm is described.Finally,it is verified that the accuracy and real-time performance of the information fusion method is significantly higher than the visual method by the real vehicle experimental platform.(3)The problem of lateral obstacle avoidance control for intelligent vehicles after detection of unknown obstacle vehicles is studied.By using double lane condition as global reference trajectory.Next the basic principle of model predictive control and the establishment of the predicted model are introduced in detail,and design of trajectory tracking controller by setting the constraint conditions and optimizing the algorithm.Then the trajectory re-planning controller is designed by adding the obstacle avoidance equation and the five order polynomial in the algorithm.Finally,the active obstacle avoidance control of the unknown dynamic and static obstacle vehicle is verified by a joint simulation platform experiment.