Research On Key Technology Of Indoor SLAM Mapping Based On TOF Lidar

With the continuous development of artificial intelligence,sensor technology and computer technology,the field of robots has also been greatly developed.Mobile robot mapping is the key to robots' autonomous navigation and environmental exploration.It plays an important role in different fields such as space exploration,underwater exploration and indoor application.It has important research significance and application value.Firstly,aiming at the phenomenon that the price of 2D lidar is generally high on the market,a low-cost 2D lidar system based on Time of Flight(TOF)is proposed.Secondly,this paper studies the Simultaneous Localization and Mapping(SLAM)method based on 2D lidar.In the mapping,the gradient descent method is sensitive to the initial value,easy to fall into the local minimum value,non-convergence,etc.,using the Correlative Scan Matching to obtain the relative pose,and participate in the second time as the initial value of the nonlinear optimization problem.Among them,the method based on submaps scan matching is adopted to reduce the accumulated error in the process of mapping.The Gauss-Newton method is used to solve the nonlinear optimization problem of scan matching to improve the robustness and accuracy of matching.Aiming at the problem that the amount of information of single-frame data in the loop closure detection is less and the error closed loop is generated in a similar environment,a loop closure detection method based on key frame matching is proposed.Among them,the matching frame is formed by the key frame and the current frame,the amount of information is increased,and the correct rate of the loop closure detection is improved.The bicubic interpolation algorithm is used to optimize the map to obtain a map with better continuity,which improves the accuracy of scan matching.Finally,the Sparse Pose Adjustment(SPA)back-end optimization method is studied.The constraints generated by loop closure detection are used to adjust the pose of all submaps and reduce the accumulated error of the front end.This paper verifies the proposed method by building a mobile experimental platform.The experimental results in the actual scene mapping show that the proposed algorithm effectively avoids matching errors during the mapping process,reduces errors in the process of mapping,and constructs a map consistent with the environment.