Research On SLAM Technologies For Intelligent Wheelchair System Based On Lidar And UWB

In recent years,the intensification of population aging leads to rapid expansion of the elderly population.How to meet the elderly and physically disabled people’s needs for free indoor activities and outdoor activities is currently a issue that needs to be solved urgently.Traditional electric wheelchairs are difficult for users to operate and are prone to collisions during travel.To solve the above problems,the thesis is focused on the problem of building indoor environment maps and autonomous navigation for intelligent wheelchairs.The content of the thesis can be expressed into the following three aspects:First,Lidar Simultaneous Localization and Mapping(SLAM)is used to construct the environmental map for the wheelchair,and the data collected by the Lidar and the registration process are optimized.Perform statistical filtering,down-sampling and other pre-processing on the raw data collected by lidar to obtain a point cloud set with fewer outliers and a complete point cloud structure with fewer data points.The Interactive Closest Points(ICP)algorithm is used to register the point cloud to estimate the pose change of the wheelchair.The Fast Point Feature Histogram(FPFH)feature of the point cloud is first extracted for the coarse registration,and then the ICP algorithm is used for the accurate registration,which improves the problem of registration failure caused by too large initial position difference during the registration process.Kd-tree is used to speed up the registration speed of the point cloud.Secondly,the Ultra Wide Band(UWB)module is used to solve the global positioning problem of the wheelchair.Aiming at the problem that the three circles constructed by the TOA-based three-side positioning method in practical applications do not intersect at one point,which leads to the problem that the equation cannot be solved,an improved triangle centroid method is used for positioning.When the UWB base station is on the same plane but the height between the base station and the ground is unknown,the traditional TOA positioning method cannot be solved.For this scenario,a multilateral mobile positioning method based on the odometer is proposed for positioning.After MATLAB simulation and actual verification of the wheelchair system,the algorithm can control the positioning error within 20 cm.In order to further improve the positioning accuracy during the movement,UWB module is combined with data collected by odometer and the extended Kalman filter is used to correct the positioning results of the traditional TOA positioning algorithm,which reduces the positioning error caused by the movement of the wheelchair.Finally,the motion control module and environment perception module of the intelligent wheelchair are built,and mapping experiments are carried out in the real environment.Aiming at the problem that the traditional A* algorithm does not consider the size of the wheelchair itself when looking for a path,an improvement is proposed and its effectiveness is verified by experiments.The algorithm draws on the idea of image expansion and constructs a neighborhood matrix with the wheelchair as the center.The size of the matrix is the width of the wheelchair divided by the grid size.When judging whether a point in the grid map is passable,it is necessary to ensure that all points in the neighborhood centered on this point are non-obstructive points.The algorithm improves the problem of the wheelchair driving close to obstacles and improves the safety of the wheelchair’s automatic driving.The mapping,localization and navigation experiments of the wheelchair were conducted in the real environment,and the results demonstrated the feasibility and effectiveness of proposed methods.