基于NAO机器人目标识别与定位算法

针对目前应用最广的新型仿人机器人NAO平台,和其官网发布的目标识别算法原理,提出一种改进的识别方法,降低光照对目标识别的影响,缩短机器人识别、定位和跟踪整个过程闭环操作过程的反应时间。其中搜索路径采用原地查找、走动和旋转查找三部曲实现360度全视野搜索;目标颜色识别基于YUV颜色空间进行;目标跟踪采用目标始终在视野中心算法实现;同时分析第四代NAO机器人的硬件参数设计头关节垂直偏角与目标的距离计算公式,用于目标快速定位。

Target Recognition and Localization Algorithm based on NAO Robot

According to a new type of humanoid robot and the most widely used NAO platform,an improved identification method is proposed. It reduces the effect of light on target recognition,shortens the reaction time of robot identification,localization and tracking which is a whole closed-loop operation process. The search process uses a trilogy including in-situ search,walking and rotating lookupto fufill a 360° full visual tracking. Target recognition is based on YUV color space. Target tracking is implemented with the algorithm that always keeps the target in the center of the vision field. The NAO v4 Head Picth joint angle and the distance calculation formula are also analyzed to achieve fast localization.