Research On The Positioning Technology Of AGV Guideposts Based On Lidar

With the transformation and upgrading of our country's manufacturing industry,Automated Guided Vehicle(AGV)has become more and more widely used in logistics and warehousing,automobile assembly lines,mechanical processing and many other places.In the research of AGV related technology,positioning technology is one of the core researches,and it is the first step for AGV to realize automatic guidance.Positioning technology refers to the process in which the AGV uses optical,electromagnetic and other sensors to perceive the external environment and calculate its own posture in the environment.Aiming at the low positioning accuracy of the triangle and three-side positioning algorithms in traditional lidar positioning,this thesis uses the multi-signature positioning algorithm to carry out the following research:(1)The overall design of the AGV lidar navigation system platform was carried out according to the functional requirements.Install corresponding control units and sensors on the vehicle body to build a complete AGV experimental platform;establish the coordinate system of the Automated Guided Vehicle,and derive the odometer model and the lidar sensor model of the single-wheel forklift.(2)Preprocess the lidar scan data,design the reflection intensity filter function to extract the reflector data in the environment,and use the least square method and distance compensation method to fit the center of the cylindrical reflector;Triangular matching and a dynamic matching strategy based on historical poses are proposed,which is used to match the local coordinates of the scanning reflector with the global coordinates of the reflector.(3)Use multiple landmarks for redundant measurement,improve positioning accuracy through adjustment,and use an improved iterative search multiple landmark positioning algorithm based on the least square method to solve the pose of AGV and establish a residual analysis model to improve positioning accuracy and effectiveness of evaluating positioning;The weighted smoothing algorithm is used to fusion filter the positioning results of the lidar and the odometer pose estimation to smooth the influence of Gaussian noise in the positioning.(4)The method of using lidar to construct a reflector map is studied.Based on the known reflector map to complete its own positioning,new reflectors are continuously detected to build incremental maps;Aiming at the problem of measuring the quality of reflector layout,it is proposed to use the horizontal dilution of precision to characterize the distribution quality of the reflector in space.Finally,the positioning experiment of the lidar positioning system designed in this thesis is carried out on the AGV lidar positioning system platform.The positioning accuracy of the lidar positioning algorithm designed in this thesis and the smoothing effect of the weighted smoothing algorithm on noise are verified.