Environment Perception And Localization Of Mobile Robots Based On Multi-sensor Fusion

As an important branch of intelligent robots,mobile robots have a wide range of application scenarios.In recent years,they have attracted much attention,and mobile robot technology has also developed rapidly.Robot real-time positioning and environment perception are two key technologies to realize autonomous navigation of mobile robots,and they have been widely studied.At present,the mainstream method adopts the simultaneous localization and mapping(SLAM)technology,which uses the data collected by the sensors carried by the robot itself to perform pose estimation and map construction.In order to realize the positioning and mapping of mobile robots in unknown environments,this paper proposes a multi-sensor fusion SLAM scheme to improve the reliability and stability of the robot in the working process.Aiming at the problem of mobile robot positioning and mapping,this paper conducts indepth research on the multi-sensor fusion positioning method and real-time mapping method.The main research contents are as follows:In the realization of robot localization,the basic framework of ORB-SLAM3 is improved,and a multi-sensor fusion localization method based on RGBD camera,IMU and wheel odometer is proposed.First,the mathematical models of the three sensors are established to analyze their data characteristics;then the data fusion of the IMU and the wheel odometer is realized by the EKF method,which reduces the amount of drift generated by the IMU;a new feature detection algorithm is proposed for the visual front-end.At the same time,the data obtained by EKF fusion is used as the prior information of the front-end for pose tracking,which effectively improves the work efficiency of the front-end.The data is fused to achieve the optimal estimation of the pose.In terms of robot mapping,RGB images,depth images and corresponding pose information are used to convert depth information into point cloud maps in the world coordinate system.The point clouds under the multi-frame images are spliced by the ICP algorithm to obtain a continuous point cloud map.Then convert the point cloud image into an octree map and an occupied grid map to update and store the map.In order to ensure the realtime performance of the mapping system,the constructed maps are divided into local maps and global maps.The global map is updated through key frames,which reduces map redundancy and improves mapping efficiency.Finally,a mobile robot platform is built,and the experimental verification of the positioning system and the mapping system is carried out.The experimental results show that the method proposed in this paper has good performance in terms of positioning accuracy,mapping efficiency and system stability,and effectively improve the reliability and stability of the mobile robot work.