基于双目视觉SLAM的室内机器人导航技术

Indoor Robot Navigation Technology Based on Binocular Vision SLAM

本文首先介绍了视觉SLAM定位流程,以及基于卡尔曼滤波和关键帧的SLAM算法。并通过仿真实验对两种定位算法进行了对比,实验结果表明基于关键帧的SLAM定位更加准确。在此基础上,设计了基于关键帧SLAM算法的机器人导航系统,并对该导航系统中的ORB特征提取与匹配、栅格地图构建、移动机器人路径规划等技术要点展开了分析。最后通过实验对机器人导航系统的特征点移动匹配功能、路径规划功能进行了验证。结果表明该系统能准确匹配处连续运动图像的特征点,可规划出有效的行进路径,满足了机器人精确导航定位的需求。

This paper first introduces the visual SLAM localization process,as well as the SLAM algorithm based on Kalman filtering and key frames.The two positioning algorithms are compared through simulation experiments,and the experimental results show that the SLAM positioning based on key frames is more accurate.On this basis,a robot navigation system based on key frame SLAM algorithm is designed,and the technical points such as ORB feature extraction and matching,grid map construction,and mobile robot path planning in the navigation system are analyzed.Finally,through experiments,the feature point movement matching function and path planning function of the robot navigation system are verified.The results show that the system can accurately match the feature points of continuous moving images,can plan an effective travel path,and meet the needs of precise navigation and positioning of robots.