| As the standard of living continues to improve, it is higher demand for intelligent robots. Mobile robots in an unknown environment complete autonomous navigation and three-dimensional modeling tasks, is a key issue in the development of intelligent mobile robot. SLAM(simultaneous localization and mapping) for robot navigation and modeling of three-dimensional environment to provide a solution. Miniaturization of robots to meet today's market demands for weight sensitive micro intelligent robot in the monocular SLAM became scholars' research.Based on the monocular SLAM as the main line, the following research:1) Feature extraction and matching for monocular SLAM problem of real-time performance is not high, compared to mainstream feature extraction algorithms such as FAST, SIFT, SURF. System based on monocular SLAM SURF feature extraction and matching systems, improve system real-time.2) For monocular SLAM system based on point features map describes the efficiency and low noise and so on, line and plane features of monocular SLAM three dimensional map building process, improve search speed and stability of the system modeling of three-dimensional space. We use a rapid line search algorithm, and its efficient extraction of two-dimensional line in the image, based on two-dimensional line matches the drawing in three dimensions line.3) Based on three-dimensional point produces a minimal sample J-Linkage of high complexity is not conducive to map application characteristics of large amount of data a monocular SLAM. Through the point character line enhancement algorithm of J-Linkage characteristics improve clustering speed planes, improves the reliability of monocular SLAM map building, reduced expression of redundant information systems in three dimensions.This paper through the introduction of the monocular SLAM map building feature line and plane, the realization of the monocular SLAM point, line and plane multi information fusion 3D environment modeling. |
PDF Full Text Request