【摘要】 针对智能环境辅助的室内移动机器人导航,提出一种具有高性能、低功耗和通用性的新型分布式视觉定位系统。为克服传感器成像噪声、环境背景混杂和光照变化等对视觉目标感知造成的干扰,在投影法彩色目标快速检测的基础上,提出结合Tsai氏标定与多项式拟合的全局目标视觉定位算法,以及一种多视角协同策略,提高了在复杂环境下全局视觉定位的准确性和可靠性。研究开发了基于TMS320DM642的嵌入式DSP智能感知节点软硬件架构,实现了实时视觉信息采集、处理与网络接口等集成。移动机器人导航控制平台的应用与实验验证了分布式视觉定位系统的定位精度等性能。更多还原

【Abstract】 A novel high-performance,low-power and generic-structured distributed visual localization system is proposed for augmenting mobile-robot navigation in intelligent environment.To overcome the influence on visual target perception caused by senor,clustering background and lighting condition changes,a projection based color target recognization method is employed.A new localization algorithm combining Tsai’s calibration and polynomial fitting is proposed,as well as a multi-camera collaborative strategy.This improves the accuracy and reliability of the global visual localization in a complex environment.Hardware and software structure of smart sensing nodes based on embedded DSP TMS320DM642 is developed.Key functions of real-time visual information acquisition,processing and network interface are integrated in the smart nodes.Application and experiments with the mobile robot navigation control platform verifies the localization accuracy as well as other favorable performances.更多还原