嵌入式全方位视觉导航器研究

Research on an embedded omnidirectional vision navigator

研制了一种基于嵌入式技术和鱼眼镜头的新型全方位视觉导航器,可控制不同种类的移动机器人平台在光照剧烈变化的室内以及强日光照射下的室外环境中完成自主导航。提出一种以改进的粒子滤波算法(MPFT)为核心的全方位视觉导航算法,将均值漂移和动态阈值调整嵌入粒子滤波器,提高了跟踪算法的效率和对光照变化的适应能力。结果表明,在不同移动载体和不同导航环境的情况下,嵌入式全方位视觉导航器具有很好的通用性、鲁棒性和实时性。由于系统集图像采集、分析和处理于一体,体积小、功耗低和通用性强,具有非常广泛的工业应用前景。

A novel omnidirectional vision navigator based on embedded technology and fish-eye lens is researched. It can control different kinds of mobile robots to finish the automatic navigation in the indoor condition with large scale illumination variation and in the outdoor environment with strong sunlight. An omnidirectional vision navigation algorithm with the core of multi-target tracking is presented. The efficiency and adaptability with illumination variation of tracking algorithm have been improved by implanting mean-shift and dynamic threshold adjustment into the particle filter. The results demonstrate that the embedded omnidirectional vision navigator has high ability of commonality,robustness and real time control with different mobile robots and navigation applications.