基于视觉SLAM和人工标记码的定位与导航方法研究

Research on Positioning and Navigation Method Based on Visual SLAM and Manual Marking Code

针对移动机器人编码里程计存在的系统性误差和非系统性误差问题,采用扩展卡尔曼滤波算法,将精准度较高的视觉里程计与编码里程计进行信息融合,以提高机器人定位精度.针对机器人直行中存在的航向角偏差,采用人工路标对机器人进行轨迹纠偏和导航.通过人工路标的快速识别,保证机器人按照既定规划路径行驶和实时导航.在室内环境下分别进行了定位实验和导航实验.定位实验表明,融合后的里程计极大地提高了机器人的定位精度,定位误差被控制在5 cm内;导航实验表明,基于人工路标的导航方法可以顺利地帮助机器人完成导航任务,从而证明了该方法的可行性和实用性.

Working in an indoor environment, the autonomous localization and navigation technology of mobile robots is the basis and key to complete advanced operations, So it has very important research significance. In view of the system and non-systematic error accumulation caused by mobile robot coded odometer positioning, this paper uses a highly accurate visual SLAM odometer and extended Kalman filter to fuse it with the coded odometer information to improve the robot. Positioning accuracy;for the problem that the robot is inevitably biased during the straight-through process, the manual tracking method is used to track and correct the trajectory of the robot. The fast recognition of the two-dimensional code can ensure that the robot travels and navigates according to the planned path in real time. Experiments show that the odometer information after fusion greatly improves the positioning accuracy of the robot;the navigation experiment is carried out in the indoor environment of complex scenes, and the navigation algorithm based on artificial landmark can smoothly help the robot to complete the navigation task, which proves the feasibility and practicality of the method.