Research On 3D Environment Perception Reconstruction And Path Planning Of AGV Based On RGB-D Camera

With the continuous improvement of industrialization and the increase of labor costs,the automation level of production and handling has become more intelligent.AGV(Automated Guided Vehicle)is an intelligent handling vehicle that integrates environmental perception,planning and decision-making,and intelligent control.One of its important research contents is the effective planning of the path of the AGV trolley to the destination in a complex environment.This article aims at the path planning of the AGV trolley in the complex workshop environment,based on the Turtlebot experimental platform,equipped with RGB-D camera Kinect,completes the research of the 3D environment reconstruction and path planning method of the AGV trolley.The specific content is as follows:(1)Real-time reconstruction of 3D environment with Kinect camera.Firstly,in the feature point extraction and matching part of the front-end visual odometer,the real-time performance of several commonly used feature point extraction algorithms and the number of extracted feature points are experimentally compared and studied,and the ORB algorithm with better real-time performance is used to extract feature points.Secondly,aiming at the mismatches in feature matching,a KNN-RANSAC combination method is proposed to eliminate mismatches.Thirdly,in order to eliminate the cumulative error generated by the front-end visual odometer,a loop detection method based on the combination of local loop and random loop is adopted.Finally,in the back-end optimization link,the robot's pose estimation data obtained by the visual odometer and loopback detection is sent to the back-end,and the G2 O algorithm is used to realize the global optimization of the pose.(2)Path planning algorithm for mobile robot.Firstly,the Dijkstra algorithm,BFS algorithm and A* algorithm in the global path planning algorithm are simulated and compared and studied,and the more efficient A* algorithm is adopted.Secondly,the DWA algorithm in the local path planning algorithm is researched,and the weight coefficient of the evaluation function that affects the efficiency of the DWA algorithm is analyzed by simulation experiments.The results show that selecting the appropriate weight coefficient can improve the efficiency of the robot in the local path planning.It provides a reference for choosing the appropriate weight coefficient of the evaluation function in practical applications.Finally,a path planning algorithm combining A*algorithm and DWA algorithm is proposed.(3)Experimental results and analysis of 3D environment reconstruction and path planning in the physical scene of mobile robots.Firstly,through the experimental platform built and based on the public data,the experiment is compared with the traditional 3D environment real-time reconstruction algorithm,the results show that the algorithm improves the positioning accuracy and real-time performance.Secondly,a three-dimensional mapping experiment was carried out in the physical environment to verify the feasibility of the entire three-dimensional environment reconstruction system.Finally,based on the ROS robot operating system,experiments are carried out on the path planning method combining the A* algorithm and the DWA algorithm in this paper,and the experimental results verify the feasibility of the algorithm.