基于粒子滤波的全方位视觉传感器实现移动机器人导航  

Omni-Directional Visual Sensor for Mobile Robot Navigation Based on Particle Filter

作　　者：李雁斌[1] 曹作良[1] 刘常杰[1] 叶声华[1] 

机构地区：[1]天津大学精密测试技术及仪器国家重点实验室,天津300072

出　　处：《传感技术学报》2009年第5期745-750,共6页Chinese Journal of Sensors and Actuators

基　　金：国家"863"计划项目(2007AA04Z229);国家科技部中芬国际科技合作项目(2006DFA12410)

摘　　要：研究了一种引导机器人沿着固定轨迹运动的实时视觉伺服系统,并提出了一种实现移动机器人导航的创新的系统架构。系统由基于鱼眼镜头的全方位视觉传感器和嵌入式图像处理单元构成,可以实时校正鱼眼镜头畸变,并利用复合的粒子滤波算法识别和跟踪自定义的航标,最终实现移动机器人的导航。实验表明,该系统能够校正鱼眼镜头畸变,实时跟踪航标并有效实现机器人的定位和导航,且结构紧凑、功耗低,容易集成到各种移动设备中实现自主导航、安防监控等功能。A real-time visual servo system is developed to steer the system along the computed path and a novel system for autonomous mobile robot navigation is presented. The system consists of an omni-directional camera based on fisheye lens and an embedded image processing unit, and it can recognize and track man-made landmarks robustly based on composite Particle Filter, and realize robot navigation finally. Experiments show that this system can recover the distortion due to fish-eye lens, track the landmarks, locate and navigate the robot effectively. Besides, this compact on board system can be easily integrated into a variety of mobile devices and appears low power, well modularity and mobility.

关 键 词：视觉传感器 目标跟踪 粒子滤波 数字信号处理器 

分 类 号：TP242.6[自动化与计算机技术—检测技术与自动化装置] TP13[自动化与计算机技术—控制科学与工程]