Simultaneous Mapping And Construction Of Environmental Features With Single Camera

With the development of industrial technology and the rapid development of computer speed,mobile robot has been developed rapidly.Mobile rabot involves various fields,include intelligent control technology,computer technology,pattern recognition and artificial intelligence,etc the rapid development of these subjects also contributed to the development of mobile robot.At the same time the laser range finder, sonar, radar, GPS and other advanced sensor development in the mobile robot is also widely used. Mobile robot attracted the attention Chinese and foreign scholars, the major colleges and universities conducted on the application of mobile robot research, achieved a lot.Based on visual information for autonomous mobile robot challenge is how to improve the robustness of the visual system to adapt to environmental changes, how to get from a single camera image data, processed to restore the depth and accurate information to determine the robot's own pose, How to solve the problem of real-time process to achieve the robot's own movement speed and flexibility, this paper focuses on aspects of the research, design the corresponding method.In this paper, the visual robot localization and map building for the research, design of a monocular vision mobile robot simultaneous localization and map building method of the environment. In this paper, using Harris-SIFT (Scale Invariant Feature Transform) feature extraction algorithm for extracting environment features, while building environmental characteristics database, using RANSAC (Random Sample Consensus) to find an affine matrix, get real-time image feature points and feature points in the corresponding features of the database point the relative position, combined with extended Kalman filter to achieve localization and construction of environmental features. The article first describes the visual problems, then introduces the key technologies of visual localization and Research situation about visual localization.On this basis leads to the focus of this paper. Next, this paper, the key technologies are described in detail, including the Harris-SIFT feature extraction and matching, the basic principle of Kalman filtering, extended Kalman filter localization of the motion model, the measurement equation, prediction and parameter update feature points. The experiments validate the location and environmental characteristics of construction feasibility of the method, analysis method of problem...