Autonomous Exploration And Mapping Of Robots In Unknown Indoor Environments

In recent years,robots have been rapidly integrated into people's daily lives.In order for the robot to navigate autonomously through unexplored indoor space,an accurate map must be provided.But at present,the drawing of the robot still depends on manual control,which leads to too much drawing accuracy and time cost,and requires the operator's technology to move,the hope is that robots will be able to move autonomously in unknown environments at their will.Therefore,it is very important to let the robot construct the map and improve the efficiency of autonomous exploration.Aiming at the problem that the robot can not construct map autonomously in the unknown environment and can only construct map of small scene,and most of the existing autonomous exploration is sensing the environment by Lidar isometric sensor,then a two-dimensional grid map of the environment is created.In this paper,the autonomous mapping algorithm of robot in unknown environment is taken as the research topic,and the autonomous mapping principle based on RRT algorithm of fast random search tree and two different SLAM algorithms are mainly studied,this article does the following.In the self-exploration principle section.This paper mainly studies the structure and principle of the basic fast random search tree(RRT)algorithm,on the basis of which,the RRT algorithm is improved,which is migrated from the path planning algorithm to the field of boundary detection,and a multi-step search strategy is proposed,the proposed strategy consists of three parts.First,we construct an initial two-dimensional raster map using lidar data,and use local RRT detection to detect the boundary points around the robot,a global RRT with adaptive step size is used to detect the remote boundary points of the constructed map.The boundary points are then filtered and clustered to reduce the total number and computational load.Finally,the boundary point evaluation function is constructed by three factors: the information gain of the robot in the process of autonomous exploration,the cost of navigation and the accuracy of the robot self-localization,after selecting the boundary point with the highest score as the best boundary point,the closed-loop control is used to guide the robot to the best position until the map is constructed.In the aspect of map building with SLAM,the paper first analyzes the two main methods of map building based on particle filter and graph optimization,and then analyzes the principles of location and composition,advantages and disadvantages of the two typical algorithms,gmapping and cartographer.Finally,aiming at the validity of the above algorithms,the simulation experiment environment is built and the results are analyzed after all the experiments.