SLAM And Path Planning Of Autonomous Robots In Unknown Environments

In recent years,autonomous robots have gradually entered people's daily life,helping people to complete some tasks that people can or can't do well,which is convenient for people's life.In recent years,the navigation system of autonomous robot is the research object all over the world,which plays an important role in both industrial robot and service robot.In such an era,this paper aims to develop a navigation system based on ROS robot system,complete the basic map construction and path planning of robot,and design a multi robot system to complete the path coordination between multi robots.According to the overall analysis of the robot system,the robot is divided into interaction layer,function decision layer and hardware execution layer,and the tasks of each layer are analyzed in detail.By constructing the motion model of the robot,the mathematical foundation for the algorithm design and robot control is established.The communication system from the control end to the executive end is designed to ensure the autonomous navigation function of the robot.Aiming at the SLAM algorithm design of robot,this paper analyzes the advantages and disadvantages of traditional SLAM algorithm,and finally improves it on the basis of Hector slam.A time tag is added before the coordinate transformation of the point cloud data collected by lidar,and then the scanning points are mapped to the global coordinate system according to the robot pose corresponding to each scanning point,and then combined with the odometer information for further pose estimation,which improves the matching degree with the real environment and strengthens the authenticity of the constructed map.For the design of path planning algorithm,a fuzzy Q algorithm designed combines fuzzy control and Q-learning algorithm,and designs the state action pair,Q matrix,reward function and action strategy.Three distances and one angle are used as the input of fuzzy control,and one action is used as the output(there are seven kinds of actions),and a fuzzy rule table is designed for the control object.A variable step update strategy is designed to speed up the iterative speed of the algorithm.In the choice of action strategy,a new balance algorithm is designed to balance the contradiction between action execution and action exploration.Aiming at the design of multi robot system,this paper selects the hybrid control mode,analyzes the problems that may appear in the process of multi robot moving,and improves Dijkstra algorithm for multi robot path coordination,solves the conflict and deadlock problems between robots.In this paper,we use STDR and Turtlebot2 to simulate and experiment the designed algorithm respectively,and verify that the algorithm has good feasibility and efficiency.