项目名称: 面向模具保护的无标定视觉伺服控制方法研究
项目编号: No.61202203
项目类型: 青年科学基金项目
立项/批准年度: 2013
项目学科: 计算机科学学科
项目作者: 邢科新
作者单位: 浙江工业大学
项目金额: 24万元
中文摘要: 课题针对当前的模具保护器报警后需要人工干预处理残留物,无法实现全自动化无人值守的缺点。将伺服机械手引入模具保护系统,提出利用机器人无标定视觉伺服技术进行模具残留物的实时检测、定位和抓取的新思路。从提高模具残留物目标定位和跟踪的鲁棒性与实时性角度出发,研究抓取视觉对象的降维局部特征提取算法。总结现有的雅可比矩阵在线迭代估计算法的共性,设计统一的算法框架,并利用回声神经网络的自适应性和计算的高效率特性,设计图像雅可比伪逆矩阵的在线估计算法及其运动学视觉伺服控制算法。并将上述基于运动学的视觉伺服控制算法扩展到动力学视觉伺服控制中,同时考虑系统动力学模型及外部扰动存在的不确定性,引入扩展状态观测器对其不确定特性或扰动进行补偿,设计具有鲁棒特性的自适应动力学视觉伺服控制方案。本课题既面向视觉伺服控制技术的理论前沿,又推动视觉伺服控制技术在模具保护行业的实际应用,具有重要的理论和现实意义。
中文关键词: 模具保护器;视觉伺服;特征匹配;自适应;
英文摘要: Now the remain on the mold need to be removed artificially when the mold monitor and protection system takes alarm. This artificial process leads to low efficiency in production. To implement full automation and no-man on duty during production processes, the servo manipulator and uncalibrated visual servo control technology will be adopted to implement inspection, location and grabbing of the foreign substances. In order to improve the real-time characteristic and robustness of the object tracking, this work plans to research the dimension reduction method of the local feature for the tracking object. At the same time, to develop the visual servo control algorithm based on the pseudo-inverse matrix of the image Jacobian, which is estimated online by the high efficiency and adaptive echo state network. Considering the uncertainty of the manipulator's dynamic and the external disturbance, the study integrates the preliminary methods to develop the robust and adaptive dynamic visual tracking control strategy. This work explores the advanced visual servo control algorithm and promotes its actual application in mold protection area. This is a research work with positive academic and practical significance.
英文关键词: Mold Protection;Visual Servo;Feature Matching;Adaptive;