Research On Visual Localization And Path Planning Of Indoor Mobile Robot

Mobile robot is a kind of comprehensive system that combines the functions of information collection,behavior decision-making,motion control and so on.It is active in many fields such as housekeeping,security,disaster relief,detection and so on.It plays an increasingly important role in our daily life and has huge potential demand.With the continuous expansion of mobile robot application scenarios,the traditional robots that rely on a single sensor are difficult to adapt to the complex and changeable indoor environment,and the accuracy,adaptability and safety of autonomous positioning are insufficient.Therefore,it is of great significance to improve and optimize the autonomous positioning and path planning process of mobile robots.This paper constructs an indoor autonomous navigation system based on ROS,which realizes indoor high-precision positioning and navigation obstacle avoidance by integrating monocular camera and inertial sensor data.Aiming at the narrow indoor passage,an omnidirectional mobile robot chassis is built.Taking the whole robot as the object,the kinematics model is established,and the relationship between the linear velocity,angular velocity and wheel linear velocity of the robot is analyzed;In order to obtain the precise pose of the robot,the visual inertial odometer is constructed by the close coupling of monocular camera and IMU sensor data,and the indoor two-dimensional map is constructed by taking the pose estimated by the odometer and the two-dimensional point cloud of single line lidar as the input;Because the efficiency of traditional path planning algorithm is not high,a fast ARA*path planning algorithm is proposed,which uses binary sort tree-instead of linear storage structure to store node information,which improves the efficiency of path planning;Aiming at the problem of insufficient security when the robot moves,an adaptive node connection strategy is proposed.Finally,combined with the fast ARA*global path planning algorithm and DWA local path planning algorithm,the autonomous navigation and dynamic obstacle avoidance of the robot are realized.The experimental results show that the visual inertial odometer constructed in this paper can achieve indoor high-precision positioning,and the positioning accuracy is 8.6cm;The proposed fast ARA*path planning algorithm reduces the search time by 43%,and ensures the safe distance between the robot and obstacles.This paper solves the problems of low flexibility,poor robustness,and insufficient efficiency and security of path planning for the robot based on a single sensor,and provides an effective method for the application of mobile robot in complex indoor environment.