Abstract: To reduce the influence of cumulative odometer localization error, an autonomous localization approach for mobile robot based on odometer and Pan-Tilt-Zoom (PTZ) vision was presented. Interrupt S-shaped searching strategy was put forward, as well as a vision localization approach based on finite automaton. The position credibility models of odometer and PTZ vision were established according to their positioning error sources. A localization framework combining odometer and PTZ vision based on real-time position credibility was designed. An application experiment of vision localization and localization using odometer and PTZ vision was carried out. The validity and practicability of the method are confirmed.

Key words: mobile robot, interrupt S-shaped searching strategy, credibility model, Pan-Tilt-Zoom (PTZ) vision, odometer, localization

摘要： 针对机器人长距离运行时里程计定位存在累积误差问题,提出一种基于里程计和PTZ视觉的移动机器人自定位算法。提出了中断式S形搜索策略的概念,设计了基于有限自动机的视觉定位方法;分析了里程计和视觉定位误差来源,分别建立了其定位信度模型;并基于该模型建立里程计和PTZ视觉定位的框架。针对视觉定位及里程计视觉复合定位分别进行了实验,结果验证了该方法的有效性和实用性。

关键词: 移动机器人, 中断式S形搜索策略, 信度模型, PTZ视觉, 里程计, 定位