基于双目视差和主动轮廓的机器人手眼协调控制技术研究

Robotic Hand-Eye Coordination Control Based on Binocular Disparity and Active Contour

提出一种基于双目视差和主动轮廓的机器人手眼协调技术;该方法利用主动轮廓的思想动态地逼近和跟踪机器人及目标物体的外部轮廓,通过控制双目视差趋零来实现机器人靠近目标和抓取物体.首先建立了机器人手指轮廓的几何参数模型和相应的观测概率密度模型,利用CONDENSATION算法对轮廓进行动态逼近和跟踪.然后针对动态轮廓的几何特征,探讨了基于双目视差的手眼协调控制策略.最后利用这种手眼协调控制技术进行了机器人抓取圆球目标的实验.实验表明,这种方法具有对图像噪声不敏感的特点,可以在杂乱的背景环境和复杂的纹理条件下执行视觉引导的跟踪和抓取任务,具有较强的鲁棒性.

Robotic hand-eye coordination technology based on binocular disparity and active contour is proposed in this paper. The idea of active contour is used to dynamically approximate and track the external contour of the robot and the object, and the method which makes the binocular parallax to zero is proposed to enable the robot to reach the target and grasp the object. Firstly, the geometric parameter model of robot finger contour and the corresponding probability density observation model are set up, and CONDENSATION algorithm is applied to approximate and track the contour dynamically. Then based on the geometrical characteristics of approximated contours, the strategy of hand-eye coordination control based on binocular disparity is discussed. Finally, experiments for robot to grasp a ball are implemented based on the hand-eye coordination control method. The result shows that the proposed method is robust and insensitive to image noise, and can carry out tracking and grasping tasks guided by vision even under the conditions with cluttered background and complex textures.