Multi-Grasp Manipulation

Multi-Grasp Manipulation
复制标题

多抓握操作

DOI:
10.1109/robot.1995.525746
复制
发表时间:
1995
期刊:
Proceedings of 1995 IEEE International Conference on Robotics and Automation
影响因子:
--
通讯作者:
O. Khatib
O. Khatib
中科院分区:
--
文献类型:
--
作者:
David Williams;O. Khatib

文献摘要

被引文献

相似文献

本视频展示了在同时控制内力的多臂操作方面的最新成果。该方法基于两个基本概念的整合:{增广物体}和{虚拟连杆}。虚拟连杆[l]表征内力,而增广物体[2]描述系统的闭链动力学。利用这些概念,我们开发了一个多臂系统动态行为以及作用在被操作物体上的内力的模型。这个模型已被用于开发一种控制结构,该结构使这种行为解耦,从而能够在精确控制内力的情况下进行物体操作[3]。本视频中展示的实验说明了新控制结构在由两个和三个协作的PUMA 560机械手执行的各种柔顺运动任务中的有效性。这种类型的控制需要精确施加关节扭矩,这对于当前的工业机械手来说是非常难以实现的。视频的第二部分展示了一种新的力控制方法所能带来的显著改进。在这种实现方式中,机械手的腕力传感器被用于解析与力控制子空间相对应的关节扭矩。然后,这些扭矩被用作基于关节致动器/传动模型的局部关节反馈控制器的输入。
This video presents recent results in multi-arm manipulation with simultaneous control of internal forces. The approach is based on the integration of two basic concepts: The {augmented object} and the {virtual linkage}. The virtual linkage [l] characterizes internal forces, while the augmented object [2] describes the system’s closed-chain dynamics. Using these concepts, we have developed a model of the dynamic behavior of both the multi-arm system and the internal forces acting on the manipulated object. This model has been used to develop a control structure that decouples this behavior, allowing object manipulation with accurate control of internal forces [3]. The experiments shown in this video illustra,te the effectiveness of the new control structure in various compliant motion tasks performed by two and three cooperating PUMA 560 manipulators. This type of control requires accurate application of joint torques, which is very difficult to achieve with current industrial manipulators. The second part of the video shows the significant improvement that can be obtained with a new force control methodology. In this implementation, the manipulator’s wrist force sensor is used to resolve joint torques corresponding to the force control subspace. These torques are then used as input for local joint feedback controllers based on the joint actuator/transmission models.