Robot Synchronization Positioning And Map Construction Based On Roof Features

Simultaneously Localization and Mapping (SLAM) has been a key component of robotics in the last twenty years. There are plenty of research results in SLAM under known environment. However, robot needs to perceive the environment through sensor system since the robot has little information in unknown environment.Due to the reliability and low price of visual sensors, SLAM on visual sensors has become the focus of research. In the paper, a single camera was used as visual sensor. The paper tried to implement SLAM algorithm based on traditional Extended Kalman Filter (EKF) framework and the single camera mounted on the robot. The camera used features extracted from ceiling, since features in ceiling are simple and steady.The paper constructed 3D points from several matched features using Structure From Motion (SFM) technology. Because that error increases when robot moves, EKF framework is used to filter noise. A navigation experiment in small range indoor environment was processed after implementing the EKF-Based SLAM.The paper implemented mono-vision and EKF based SLAM. A system framework was proposed to solve SLAM. Each module of the system was implemented dependently and integrated together in the end. Experiment results demonstrate the validity of the EKF-SLAM algorithm.