Research On Mobile Robot's Navigation Based On RGB-D Camera

With the development of science and technology,mobile robots are widely used in military,industrial,and civilian fields,and play an increasingly important role in daily life.As an important embodiment of intelligent mobile robots,autonomous navigation technology has gradually become a hot topic of research.Simultaneous localization and mapping(SLAM)and path planning are key technologies for autonomous navigation research.RGB-D cameras have attracted researchers' attention in recent years because they can directly obtain the depth and image information of environment.Therefore,it is of great theoretical significance and practical application value to study the autonomous navigation of mobile robots.Firstly,through the detailed analysis of navigation-related technologies and sensors' characteristics,RGB-D camera is selected as navigation sensors.The design of mobile robot's navigation using RGB-D camera on robot operating system(ROS)is completed.Secondly,aiming at the problem that the map is not consistent which caused by accumulation of errors in Visual Simultaneous Location And Mapping(V-SLAM),a visual SLAM system based on improved closed-loop detection algorithm is proposed.In order to reduce the cumulative error caused by the long operation of mobile robots,this paper introduces an improved closed-loop detection algorithm.By improving the similarity score function,it reduced the perceived ambiguity and finally made the recognition rate of closed-loop higher.In the front end,Using RGB-D Camera to simultaneously acquire images and depth information,and using Oriented FAST and Rotated BRIEF(ORB)features with low computational complexity,can effectively reduce the calculation of pose estimation for visual odometers.At the same time,the unstructured iterative optimization of the calculated pose by Perspective-n-Point(PnP)is achieved through general graph optimization(g2o),which improves pose estimation accuracy of the visual odometer.Finally,a graph optimization method based on Bundle Adjustment(BA)is used to optimize the pose and signposts in the back-end.The final experiment shows that the system can meet the real-time requirements,and can obtain a more accurate pose estimation.Thirdly,in the research of path planning of mobile robots,a path planning algorithm for mobile robot based on hybrid ant colony algorithm is proposed,which effectively improves the operating efficiency.At the same time,it avoids falling into the local optimum.Combining the target gravitational force generated by the artificial potential field,the algorithm constructs comprehensive heuristic information;which makes ants look for the path along the target to improve search efficiency.Then,the pheromone is updated by adopting the wolves' distribution principle to avoid falling into a local optimum.At last,the path optimization algorithm is used to optimize the planning path so that the planning path is smoother and the path can be further shortened.Finally,a mobile robot navigation system based on RGB-D camera is designed and implemented on the ROS platform.Through related experiments,it is proved that the proposed mobile robot navigation system is feasible and stable.