Research On Simultaneous Localization And Mapping Of Indoor Mobile Robot Based On KinectV2

Simultaneous localization and mapping(SLAM)is a process in which the robot acquires environmental data through sensors to build environmental maps and estimate its position and posture at the same time.With the increasing application of mobile robots and the complexity of application scenarios,the positioning of mobile robots in partially unknown semi-structured environments and completely unknown unstructured environments has become an urgent requirement.SLAM has become a key technology for solving autonomous navigation of mobile robots..This paper mainly analyzes the visual SLAM implementation method based on depth sensor,including robust image feature points detection and matching,camera pose estimation of robot motion,nonlinear least squares optimization and loop closure based on appearance.In order to slove the problems of low precision and poor robustness in the literature visual SLAM method,this paper proposes an improved RGB-D SLAM algorithm based on the existing 3D map construction algorithm.The direction and scale invariance information is added in the key points of the detection through the gray centroid method and the image pyramid,so that the camera can still estimate the pose trajectory under the condition that the mobile robot is shaken and rotated;the hybrid iterative closest point(ICP)and perspective are adopted.Pose estimation algorithm for perspective-n-point(PnP)algorithm;nonlinear least squares optimization model for state estimation by bundle adjustment(BA);through the loop closure based on the bag of words model and the term frequency-inverse document frequency algorithm,the influence of cumulative error is effectively eliminated,and the relocation under loss tracking condition is realized,and the accuracy and robustness are improved.Through the simulation experiment on public dataset,the pose estimation and point cloud map construction of mobile robots are realized and the absolute trajectory error and relative pose error are analyzed.The experimental results show that the improved SLAM method has a certain improvement on the accuracy and robustness of pose estimation and mapping,and can also realize real-time simultaneous localization and mapping in large scenes.Finally,an experimental platform was built on the wheeled robot with the depth camera KinectV2 in the laboratory to carry out the localization and mapping experiments in the indoor environment.The actully experimental results show that the improved system can acquire the robot motion trajectory in the indoor*unstructured environment in real time,and further verify the performance and reliability of the improved SLAM method.