Stability analysis and optimum position planning for dexterous grasping of multiple objects

多指机器人手具有灵巧地抓取和操作任何形状物体的潜力。在这个研究项目中，我研究了多目标的抓取稳定性和在此基础上的最优抓取规划问题。在平面情况下，分析了手指与物体具有滑动和滚动接触的多个物体的抓取。通过构造抓取刚度矩阵的偏导数，解析地得到了抓取参数对抓取稳定性的影响。在空间情况下，不仅考虑了滑动接触和滚动接触，还考虑了接触面几何形状(度量张量、曲率和扭转)。提出了一种最优抓取规划算法，并利用单目标抓取实验装置验证了该算法的有效性。为了计及手部的转动关节和连杆长度，采用了转动刚度模型。在模型的基础上，解析推导出了抓取刚度矩阵。

Multifingered robot hands have the potential to dexterously grasp and manipulate any shaped objects. In this research project, I have investigated the grasp stability of multiple objects and the optimum grasp planning problem based on the analysis. In the planar case, grasps of multiple objects with sliding and rolling contacts between fingers and objects were analyzed. The effects of grasp parameters to the grasp stability were analytically obtained by formulating the partial derivative of the grasp stiffness matrix. In the spatial case, not only the sliding and rolling contacts but also contact surface geometry (metric tensor, curvature and torsion) were considered. An optimum grasp planning algorithm was proposed and its effectiveness was demonstrated by using experimental devices for grasping a single object. In order to include the revolute joints and link lengths of the hands, rotational stiffness model was used. The grasp stiffness matrix was analytically derived on the model.