Research On Indoor Mobile Robot Navigation Based On Lidar

In recent years,mobile robot navigation technology is widely used in field exploration,autonomous driving,smart home and other fields.Simultaneous Localization and Mapping and path planning,as the key links in the navigation task,are gradually becoming the research hotspot of frontier science and technology.In this paper,we studied slam and autonomous navigation of mobile robot based on lidar,and realized autonomous navigation of mobile robot in dynamic environment.The main work of this paper is as follows:(1)Aiming at the problem that the accumulated error of odometer in laser Simultaneous Localization and Mapping,we calibrate the odometer with scanning matching and least square method;we compare the performance of ICP and PL-ICP in laser registration,and apply them to gmapping.Then,we use linear interpolation method to correct the laser motion distortion.Finally,experiments verify the accuracy and robustness of the method our proposed.(2)Aiming at the problem of path planning for mobile robot,we adopted global and local path planning.Firstly,in global path planning,we introduced the improved Jump Point Search based on Astar,and on this basis,we use block operation to speed up pruning rules,which can quickly identify multiple jump points in a single time.At the same time,aiming at the invalid search of Jump Point Search on symmetric nodes,we designed a heuristic function to destroy symmetry,which can avoid the symmetric nodes.Secondly,in local path planning,we described the objective function and trajectory optimization method of Timed Elastic Bands,and aiming at the local optimization problem by nonlinear optimization,we sampled the point of local trajectory through Probabilistic Roadmap to improve the optimization problem,and the trajectory can be improved by combining with the concept of homologous class to reduce the optimization of similar paths.(3)Aiming at the navigation problem in dynamic environment,we introduced a method of dynamic obstacle detection based on cost map.In this method,the running average filter is used to detect the grid value of dynamic obstacles on the cost map.And then,we used the Kalman filter is to track the obstacles on the map.At last,we combine it with the local path planner to realize the autonomous navigation on the multi-pedestrian dynamic environment.