基于Adept机器人的视觉伺服控制系统

A visual servoing system based on Adept robot

中文关键词  机器人  图像矩  视觉伺服  目标跟踪

英文关键词  robot  image moments  visual servoing  object tracking

中文摘要

      机器人视觉伺服控制在理论和应用等方面还有许多问题需要研究, 例如特征选择、系统标定和伺服控制算法等. 针对Adept机器人, 提出了一种简单快速的不需要精确标定摄像机内外部参数的摄像机标定方法, 完成了从被观测物体表面所在的视觉平面坐标系到机器人基坐标系的坐标变换. 使用图像的全局特征, 即图像矩特征进行伺服跟踪；利用所推导的图像雅可比矩阵, 设计了由图像反馈与目标运动自适应补偿组成的视觉伺服控制器. 将算法对静态目标的定位实验进行了验证, 然后又将其应用到移动目标的跟踪上, 通过调节和优选控制参数, 实现了稳定的伺服跟踪和抓取. 实验结果表明采用图像矩作为图像特征能够避免复杂的特征匹配过程, 并且能够获得较好的跟踪精度.

英文摘要

      For robot visual servoing, the influences of characteristics extraction, system calibration and control algorithm on tracking speed and precision are important. A simple and fast camera calibration algorithm is proposed, by which the internal and external parameters of camera need not be calibrated precisely. The image moments that are global image descriptors are used for servoing tracking. With the moment-based image Jacobian matrix, the visual servoing controller composed of image-based feedback and adaptive motion compensation is designed to track the moving target. The validity of the control method is first validated in the experiment where the camera is repositioned with respect to a static target. This work is extended to the tracking of moving objects. By adjusting and optimizing the control parameters, the stable servoing tracking can be realized. The experiment results show that using image moment as image characteristics can avoid the complex characteristics matching process, and acquire a satisfactory tracking accuracy.