Obstacle Detection And Tracking Research And Optimization Implementation Of Mobile Robot Embedded Platform

With the development of unmanned ground vehicles(UGV),more and more intelligent mobile robots come into being.All kinds of mobile robots in the vastly different environment of automatic driving to meet the needs of all aspects of people.Environmental perception is one of the important topics in mobile robot research,mobile robots running on structured or unstructured roads need to have obstacle detection capabilities.At the same time,the demand for low cost,low power consumption,and miniaturized mobile robots is increasing.Based on the above requirements,this paper starts from three aspects.First,build a mobile robot platform based on an embedded system.The platform needs to combine the characteristics of embedded system and micro mobile robot,analyze the system requirements,complete the equipment selection task.According to the embedded micro-controller design to achieve the movement system and embedded linux system design to achieve environment perception system,respectively for mobile robot control and data fusion environment perception functions.Then,based on the constructed mobile robot platform,the obstacle detection algorithm is designed.Obstacle detection is divided into positive and negative obstacles detection.Because of the physical characteristics of negative obstacles,it is difficult to be detection.This paper focuses on the detection of positive obstacles at the same time.By changing the single-line laser radar installation to detect negative obstacles,due to single-line lidar coverage is limited,the use of target tracking method to continuously capture obstacles,to obtain better detection results.Finally,due to the large amount of computational complexity of the target tracking algorithm used to bring real-time challenges,This paper presents the use of single-line laser radar to track the movement of obstacles to determine the direction of movement of obstacles,extract the edge of the obstacles and color characteristics to determine the general position of obstacles.In the calculation of edge and histogram statistics,the need to use i.MX6Q hardware platform GPU acceleration module,through experiments,can effectively speed up the implementation of the algorithm.And finally completed the work on the obstacle detection tracking optimization.In the future development of mobile robots,new demand will continue to produce.This paper completes the design of embedded system of mobile robot,achieves the goal of obstacle detection,and puts forward the key points of performance optimization,which is of great significance for the future research of mobile robot.