Research On Unmanned Environment Perception System And Obstacle Detection

In recent years,with the continuous growth of private car ownership,vehicle exhaust emissions are also increasing year by year,causing serious environmental pollution problems.In addition,traffic accidents occur frequently around the world,and countless families have lost their lives.Unmanned driving technology can improve fuel utilization rate and ensure driving safety,thus alleviating road traffic safety and environmental pollution problems.Environmental perception system is the foundation of unmanned driving system,and also the basic guarantee of safety and intelligence of unmanned vehicle.Obstacle detection technology is the core of environmental awareness system,and also the basic condition for unmanned vehicles to achieve automatic driving.This paper studies the key technologies of unmanned environment sensing system,including system design,sensor testing,obstacle detection and so on.The main work of the paper is as follows:1.The overall scheme design of the unmanned-driving environment perception system is completed,including the sensor configuration,communication architecture construction,data processing,fault diagnosis and other working modes.The system takes engineering machine as the core processing unit,and equipped with two millimeter wave radar,two industrial cameras,one IBEO lidar,one Velodyne 64 lidar,one pair of RSDS millimeter wave radar and one GPS inertial navigation system.2.In this paper,the method of sensor calibration is studied,and the interface testing and performance testing of the main sensors are completed.Using the C ++ programming language to complete interface testing and performance testing of millimeter-wave radar and IBEO radar.Ability to directly read the bottom data of the sensor and extract the required information according to its data format,improving the data application effect.3.An optimization algorithm for obstacle detection of 3D lidar based on DBSCAN is proposed,and complete experimental verification and analysis.Aiming at the problem of poor range adaptability of traditional DBSCAN algorithm in data processing of lidar,an improved method is proposed.According to the distribution characteristics of lidar data,the fixed cluster neighborhood radius ? in the traditional DBSCAN algorithm is adjusted to the adaptive clustering parameter which changes with the target distance.The experimental results show that the clustering effect of the optimized DBSCAN algorithm is significantly improved.