基于人工免疫和图像直接反馈的视觉伺服控制

Visual Servoing Control Based on Artificial Immunity and Direct Image Feedback

传统的基于图像特征的视觉伺服控制方法通用性差,且图像特征的标记、提取与匹配过程复杂.为此,文中提出了一种基于人工免疫和图像直接反馈的平面视觉伺服控制方法.首先采用人工免疫算法求解期望图像与反馈图像之间的位置与旋转误差,然后将误差信号经视觉控制器转换成机器人空间的位姿增量信号发送给机器人控制器,从而在无需进行图像特征标记与提取的情况下完成视觉伺服闭环控制任务.实验结果表明,采用文中方法能控制机器人准确地跟踪目标物体,控制误差不超过1个像素,说明文中方法是有效和正确的.

In order to improve the universality of the traditional visual servoing control methods based on image fea- tures and to reduce the complexity of the marking, extracting and matching of image features, a novel plane visual servoing control method is proposed based on artificial immunity and direct image feedback. In this method, first, the translation and rotation errors between the reference and feedback images are obtained by employing an artificial immune algorithm. Then, the position and orientation increments transformed from the image errors via a visual controller are transmitted to the robot controller, which implements the closed-loop visual servoing control without marking and extracting any image features. Experimental results indicate that the proposed method makes the robot accurately track objectives with an error under one pixel, showing its effectiveness and correctness.