网球捡球机器人视觉导航系统设计

Design of Vision Navigation System for Tennis Picking Robot

通过视觉对网球进行识别和定位,引导捡球机器人进行网球拾取。视觉系统采用ARM contex-A8内核的图像处理系统,结合OpenCV图像处理与机器视觉库,进行开放环境下的网球识别与定位,利用RGB颜色模型的B分量和HSV颜色模型的H分量进行前景目标的提取,基于网球的圆形轮廓特征进行目标识别,通过目标区域质心的获取进行目标定位,在局部视域下进行目标跟踪。实验结果表明,该方法在复杂的开放环境下,实现了网球目标识别和定位跟踪,为捡球机器人提供了视觉导航。

The vision system of ball-picking robots recognize and locate tennis by vision and guide the robot to pick up tennis.The vision system use the ARM contex-A8 kernel image processing system,with OpenCV image processing and machine vision library,to carry out tennis recognition and location in an open environment.The B component of the RGB and the H component of the HSV color model are used to extract foreground targets,object recognition through the circular contour features of the tennis ball,object location is achieved through the acquisition of the centroid of the object area,and object tracking is performed in partial field view.The experimental results show that the method realizes tennis recognition and positioning tracking in a complex open environment,and provides visual navigation for the ball picking robot.