Modelless and grasping-forceless control by robotic fingers capable of mechanically coupled movement
Modelless and grasping-forceless control by robotic fingers capable of mechanically coupled movement
复制标题
DOI：
10.1109/iros.2010.5652829
复制
发表时间：
2010-12
期刊：
影响因子：
--
通讯作者：
T. Inoue;Daisuke Takizawa;S. Hirai
中科院分区：
文献类型：
--
作者：
T. Inoue;Daisuke Takizawa;S. Hirai
建立了5自由度机器人手的动力学方程，该手由食指和与人手相似的对置拇指组成。该系统有两个几何约束和五个速度约束过程中产生的对象操作任务。三个速度约束与机器人的关节角度相关联，并结合到建议的系统的基础上观察的事实，即食指的远端两个关节机械耦合。基于包含这些约束的运动方程，在抓取力不确定的情况下，对5自由度机械手抓取的矩形物体进行精确的姿态控制。给定的操纵任务是单指控制（食指），其中一个串行两相（STP）控制器，以前提出的修改。在实验中，我们证明了一个简单的积分控制器，通过消除STP控制器的第二阶段设计的软指操纵工作良好。在仿真和实验中，它清楚地表明，雅可比矩阵和抓取力不一定需要在控制律完成灵巧操作任务。换句话说，这意味着在任务空间控制中可以实现完全的无模型和无抓握力控制。
This paper formulates the dynamics of 5 degrees-of-freedom (DOFs) robotic hand that consists of an index finger and an opposable thumb like the human hand. This system has two geometrical constraints and five velocity constraints generated during the object manipulation task. The three of the velocity constraints are associated with the joint angles of the robot, and are combined to the proposed system on the basis of observations about the fact that the distal two joints of the index finger are mechanically coupled. Based on the equations of motion including these constraints, we perform precise orientation control of a rectangular object grasped by the 5-DOFs robotic hand in the situation that grasping forces are not defined during the manipulation. The manipulation task given is one-finger control (index finger), for which a serial two-phased (STP) controller that had previously been proposed is modified. In experiments, we demonstrate that a simple integral controller designed by eliminating the second stage of the STP controller works well in the soft-fingered manipulation. In both simulations and experiments, it is clearly indicated that Jacobian matrices and the grasping forces are not necessarily required in the control law for accomplishing dexterous manipulation tasks. In other words, it is implied that a complete modelless and grasping-forceless control can be achieved in the task-space control.