Research And Application Of Simultaneous Positioning And Map Construction Methods For Mobile Robots Facing The Internal Environment Of Buildings

Mobile robots navigation technology is a popular research topic in the field of robotics.Its main purpose is to enable the robot move to the target position and accomplish some specific tasks in the unknown environments.Therefore,the map building and the robot's real-time localization is the foundation of autonomous navigation,which is the simultaneous localization and mapping(SLAM)technology.In this paper,we study the SLAM problem of mobile robots equipped with lidar sensor in indoor environments,especially focus on RBPF-SLAM method based on Rao-Blackwellized particle filter.In order to solve some problems of the traditional RBPF-SLAM method,such as particle degeneracy and low accuracy of scan matching,several improvement measures are proposed and the experiments have been done.First of all,we design a robot platform.On the basis of this platform,a unified model of robot system is built for experimental research.Aiming at the problem that the accuracy of conventional is low,a scan matching method based on feature points is proposed.Firstly,all the lidar points are divided into several feature segment,and the feature points are extracted based on the density and distance information in each feature segment.Then during the scan matching process,more attention will be paid to the role of feature points and feature points are assigned a higher score weight.Finally,the proposed scan matching method is introduced into the SLAM algorithm to correct the particle pose.The experimental results show that the improved scan matching method can raise the accuracy of pose estimation,and the error of the generated map is smaller.As a result,it can improve the performance of the algorithm.Aiming at the problem that traditional RBPF-SLAM method might lead to particle degeneracy and particle depletion,an improved SLAM method based on regional particle swarm optimization and partial Gaussian resampling is proposed.In order to mitigate particle degeneracy and reduce the particle number,a kind of regional particle swarm optimization method is introduced to adjust the particles' proposal distribution.All particles are clustered into several regions and the weighted central position of each region is calculated.With the particle swarm optimization,the particles of each region are drived to the regional central position to keep local convergence of the particle set.To mitigate particle depletion,during the resampling process,all particles are sorted according to their weights,then only the particles whose weight is too high or too low will be processed.Then Gaussian distribution is applied to sample new particles to keep the diversity of the particle set.The experimental results prove that the improved method can use fewer particles to generate a high-precision map and reduce the running time of the algorithm.