Research On SLAM Algorithm Of Indoor Mobile Robot Based On Multi-sensor Fusion

In recent years,with the increasing demand for industrial and consumer robots,indoor mobile robot technology has also developed rapidly.The indoor environment has many dynamic obstacles.How to control the robot to navigate and avoid obstacles smoothly in a complex indoor environment is a key issue for indoor mobile robot technology.Constructing an indoor map with rich and complete information is a prerequisite for achieving good navigation effects.This paper has explored the problem of vision and lidar SLAM.In order to solve the navigation and obstacle avoidance problem of mobile robot,a SLAM method combining mult-sensor is proposed to construct fusion map,and the experiment of navigation and obstacle avoidance is carried out using the fusion map.The main work of this paper includes:（1）Firstly,the study of visual SLAM is carried out.An improved ORB-SLAM algorithm is proposed,which improves the precision of feature matching of the image.In addition,semi-dense 3D mapping thread is added to transform the sparse 3D point cloud into semi-dense 3D point cloud,thus improving the recognition ability of the algorithm to obstacles.（2）Secondly,the paper set up a mobile robot experimental platform to build map using Gmapping algorithm.At the same time,by using the depthimage＿to＿laserscan ROS package to generate two-dimensional map of obstacles.The maps are combined by bayesian estimation method and the generated fusion maps were used in robot navigation later.（3）The map server provides the original fusion map for the navigation system,and adds the static layer,obstacle layer and expansion layer into the costmap.Navigation system uses the A* path planning algorithm to carry out the global path planning and local path planning of the robot.When the robot encounters dynamic obstacles during navigation,the lidar and the depth camera perceive the dynamic obstacles and transmit the obstacle data to the navigation system.The navigation system adds the obstacle layer generated by dynamic obstacles to the static layer of costmap,and then inflates the obstacle layer to generate the inflation layer.The robot carried out navigation and obstacles avoidance in the complete costmap.In conclusion,the mult-sensor fusion algorithm makes the constructed map more complete and informative.It can detect more static obstacles and add them to the map.The experiment of robot navigation and obstacle avoidance with the multi-sensor is carried out in the fusion map.The experimental results show that the obstacle avoidance ability of the robot is greatly enhanced after fusing mult-sensors,and the safety of mobile robot indoor navigation is improved.