Research On Autonomous Navigation Algorithm Of Indoor Mobile Robot Based On Visual SLAM

Simultaneous localization and mapping(SLAM)and path planning are the key technologies to realize autonomous navigation of mobile robots.The autonomous navigation system based on visual SLAM has a wide range of applications and can comprehensively perceive the surrounding environment,which is the key research object in the field of mobile robots.Therefore,the research on the autonomous navigation algorithm of indoor mobile robot based on visual SLAM has theoretical significance and application value.The main work of this thesis as follows.This thesis describes the domestic and foreign research status of autonomous navigation methods,visual SLAM and path planning algorithms for mobile robots,on this basis,this thesis establishes the framework of the robot autonomous navigation system based on visual SLAM.Then,this thesis analyzes the principle of camera imaging,the principle of visual odometry by the feature points method,and establishes a mathematical model for mobile robot path planning.This thesis improves the visual odometer for the problem that the visual odometer accuracy is degraded due to redundant feature points and insensitivity to moving objects.In the feature extraction stage of visual odometry,this thesis divides the image into regions,and then sets up the feature points extraction threshold adaptively according to the variation coefficient of regional grey scale,next,in order to achieve uniform extraction of ORB features,uses a quadtree structure to manage the feature points.Feature points on moving objects are removed by geometric constraints between feature points to reduce the interference of moving objects to the SLAM system.This thesis uses the progressive sample consensus(PROSAC)algorithm to eliminate feature mismatches quickly.The experimental results show that the localization accuracy of the SLAM system based on the improved visual odometry is effectively improved.This thesis researches map construction algorithms in SLAM systems,combines image data collected by RGB-D camera to construct dense point cloud maps and grid maps.In the global path planning of mobile robots,rapidly exploring random tree(RRT)algorithm has some shortcomings,such as large randomness and low path quality,this thesis proposes a path planning algorithm based on orientation information strategy RRT,by introduced direction variable and heuristic sampling strategy to guide the nodes towards the target extension,the algorithm reduces the solution time consuming and enhances the ability of the algorithm to jump out of the local optimum,and improved the path quality by smoothing the paths using cubic B spline curves.The experimental results demonstrate that the improved path planning algorithm improves the solution efficiency,and optimizes the path quality.Finally,based on the above research,this thesis uses robot operating system(ROS)as a software platform,combined with related hardware,to complete the construction of an indoor mobile robot autonomous navigation system,then,use the built mobile robot experimental platform to conduct autonomous navigation experiments in a real indoor environment with obstacle constraints.The experimental results show that the autonomous navigation system can operate efficiently and stably,which proves the effectiveness of the algorithm proposed in this thesis.