Research On Key Technologies Of SLAM For Indoor Mobile Robots

With the development of mobile robot technology,more and more working robots can replace human beings in production and life,such as navigation robot,sweeping robot,military robot,unmanned robot,etc.For some heavy work and some harsh environment,there are threats to the safety of human life,the arrival of mobile robots has more important practical significance.As a key technology for robots to obtain their own location and surrounding environment information,the importance of SLAM is clear to see.Based on the analysis of the research status and development trend of SLAM algorithms at home and abroad,this paper improved the closed-loop detection method based on the visual sensor ORB-SLAM2,proposed the closed-loop detection improved method based on histogram matching.And run the TUM data set for practical verification.The experimental results show that while maintaining the real-time performance of SLAM system,The improved closed-loop detection method can effectively reduce the positioning error of the robot system.In order to improve the mapping effect of hollow-out area and build a more accurate and dense map with SLAM technology,Gaussian statistical linear interpolation algorithm is proposed for laser point cloud interpolation.Then,the point cloud matching process in Gmapping algorithm is improved for the interpolated laser point cloud,which reduces the computation amount of point cloud matching and improves the real-time performance.Finally,the laser raster map and visual raster map constructed after interpolation are fused with Bayesian method to obtain a more perfect map.In the fusion,the choice of strategies that are more conservative for interpolation laser is more inclined to the trust of visual raster map.To verify the improved algorithm of this paper for the validity and practicability of the hollow out regional experiment scene,this paper set up a mobile robot experimental system on the scene in the laboratory actual built figure,the experimental results show that the proposed algorithm can accurately to hollow out area built figure,and can provide more accurate location information,has a good practicability.