Research On Dynamic Path Planning Algorithm For Autonomous Navigation Mobile Robot

With the increasing dependence of modern society on intelligent mobile robots,the path planning ability of robots in completely unknown or partially unknown environments has become an important direction for researchers to break through;The main research contents include the real-time map construction of the surrounding environment and the synchronous localization(SLAM)for robot,as well as the task to complete the path planning based on the constructed map and information of location.Due to the cost and volume limitations of small intelligent mobile robots,the sensors equipped on them limit their ability to perform autonomous navigation and real-time dynamic path planning in a 3D environment.Accordingly to the above issues,the designs and improvements were presented in this dissertation:(1)Firstly built the experimental platform and simulation model of mobile robot,and proposed the distributed system architecture based on the cooperative control of upper computer and lower computer;The upper computer based on the Robot Operating System was mainly responsible for collecting the information from vision sensor and the lower computer to construct the real-time map,running the algorithms to complete the path planning according to the cost map and information of location at the same time,and issuing the robot motion instructions;the lower computer that used STM32F427 series chip as the control core ran Free RTOS real-time operating system to collect the information from analog odometer,and was responsible for receiving the movement instructions from upper computer and controlling the robot platform to move.(2)Established the global cost map based on 3D elevation model and analyzed its trafficability;Established the local map based on the 3D point cloud model with depth information collected from Kinect,and then the point cloud was thinned and ground segmented to establish the real-time local cost map.The analog odometer model was built according to the encoder feedback from the motors of the platform,and the extended Kalman filter algorithm was used to integrate the data collected from the analog odometer and the global positioning module to complete the basic location of the robot.(3)Aiming at the cost map established by the three-dimensional elevation model,the traditional D*Lite algorithm based on the 2D plane was improved,so that the costs such as the travel slope and path roughness can be considered comprehensively in the process of path planning,and the global path can be completed shorter and more safety.Aiming at the problems of the traditional artificial potential field method,such as falling into the local minimum point and the trap area easily,the direction vector and virtual target point were introduced into the algorithm.At the same time,the generation of repulsion force and the calculate method were improved to complete the dynamic local path planning effectively.(4)Transplanted the improved global and local path planning algorithms into the autonomous navigation package under the framework of ROS.After selecting the cost map with known global environment and unknown local environment as the input,the path planning experiments were carried out on the actual platform to test the dynamic path planning ability of the robot,and the feasibility of the improved algorithms were verified.