Font Size: a A A

Dynamic Obstacle Detection Based On Multi-line Lidar For Autonomous Vehicles

Posted on:2022-03-12Degree:MasterType:Thesis
Country:ChinaCandidate:D S TangFull Text:PDF
GTID:2492306485994489Subject:Mechanical engineering
Abstract/Summary:PDF Full Text Request
As a means of transportation,self-driving cars are inseparable from human beings in life and work.As one of the four core technologies of autonomous driving vehicles,environment awareness technology provides accurate road environment information for autonomous driving vehicles,and it is the key to ensure the safe driving of autonomous driving vehicles.Lidar has become the main sensor of autonomous vehicles by obtaining high-precision point cloud data.With the development of technology,major breakthroughs have been made in the field of autonomous driving vehicles,but traffic accidents are still caused by the inability to fully perceive the environment during the driving process of autonomous driving vehicles.This thesis takes the autonomous vehicle with lidar as the main sensor as the research object,and combines computer vision technology and artificial intelligence technology to study the detection and tracking of dynamic obstacles in the autonomous driving environment.Build a platform for autonomous vehicles.The ranging principle of lidar and the commonly used on-board lidar in the market are analyzed,and the RS_LIDAR_16 lidar is selected as the main sensor of the entire autonomous driving vehicle to collect the point cloud information of the campus environment and road.The pretreatment of laser point cloud is completed to reduce the number of point clouds and provide a data basis for the environment sensing module of autonomous driving vehicles.The dynamic obstacle detection technology of autonomous vehicle is studied.By analyzing the principle and basic structure module of the deep learning network model,the process of feature extraction and reverse update of the network is defined.Finally,the convolutional neural network is selected as the basic structure of the obstacle detection module of the autonomous driving vehicle in this thesis,and a single-stage obstacle detector is proposed to complete the detection of dynamic obstacles in the driving environment of the autonomous driving vehicle.By using the Joint Probability Data Association Algorithm(JPDA)to establish Association matching for obstacles at different times during the driving process of the autonomous vehicle.Then,based on the untracked Kalman filter algorithm,the pose information of the obstacles at the current moment of the autonomous vehicle is predicted and updated,and the whole tracking algorithm is verified by real-time test experiments.Finally,the tracking of dynamic obstacles in the driving environment of the autonomous vehicle is completed.
Keywords/Search Tags:Self-driving cars, LIDAR, obstacle detection, obstacle tracking
PDF Full Text Request
Related items