Research On Mobile Robot Navigation And Target Following Based On Multi-sensor

At present,robot technology is more and more applied in production and life,especially mobile robot is the most widely used,and it can be seen in many industries and fields.With the continuous development of society and the rapid progress of science and technology,the technical requirements for mobile robots are getting higher and higher.Therefore,the realization of autonomy and intelligentization of mobile robots has become the future development direction and the current research focus.For mobile robots,democracy requires its capability of autonomous navigation in unknown environments,and simultaneous localization and mapping technology is the precondition of realizing autonomous navigation,as the robot to pick up the sensor in the environment of realize own positioning of the built environment map at the same time,but only with single sensor of the mobile robot positioning accuracy,poor access to information,complete map construction problems such as inaccurate.Intelligent requires mobile robots have in the case of without human intervention's ability to perform certain tasks,target to follow is a reflection of one of the most direct,and achieve the goal to follow is the goal to get the relevant information,the premise of the commonly used method is through the following goals with the color image of the camera,but this method is easily affected by the external environment and low efficiency.To solve above problems,first of all,based on the multi-sensor combination of precision positioning of the robot,which respectively using laser point cloud and visual point cloud of robot pose estimation,and then through the way of linear interpolation about their estimate of the position for joint optimization,and by using closed loop testing and optimization to get the global consistent trajectory;Secondly,Gmapping algorithm based on particle filter is adopted to construct the map and point cloud information of lidar and depth camera is fused to ensure the accuracy and integrity of the map.Then,the commonly used global path planning A* algorithm and local path planning DWA algorithm are improved.Finally,a target following method based on pseudo-laser data is proposed,which converts the depth image acquired by the camera into the corresponding pseudo-laser data and follows the target according to the developed tracking strategy.Real environment of the experimental results show that the joint position estimation method is used to estimate the position accuracy is higher and the matching between frame error smaller;sensor information fusion and closed with the optimized global positioning accuracy is greatly increased as compared with single sensor information fusion and build map more complete,accurate;after mixing path planning can effectively avoid barriers suddenly appeared in the environment of static and dynamic obstacles and planning out the global optimal route;even in a complex environment,robot can still follow the moving target on the at the same time avoid the obstacle in the environment.