非负线性组合多指抓取动态力优化方法  

Nonnegative linear combination dynamic force optimization of multi-fingered grasp

作　　者：王新庆 陈栋金[2] 姜力[2] 

机构地区：[1]中国石油大学(华东)机电工程学院,山东青岛266580 [2]哈尔滨工业大学机电工程学院,黑龙江哈尔滨150001

出　　处：《广西大学学报（自然科学版）》2015年第5期1169-1176,共8页Journal of Guangxi University（Natural Science Edition）

基　　金：国家青年自然科学基金资助项目(51305460);中央高校基本科研业务费专项资金资助(14CX02068A)

摘　　要：为了提高多指抓取动态力优化方法的快速性,提出了一种非负线性组合求初值,结合半正定规划(SDP)算法进行动态力优化的方法。该方法根据抓取力集为凸锥的特性,首先将外力旋量分解为12个单位力向量的线性组合,然后离线计算各单位力向量对应的初始基可行解,通过线性组合得到抓取力优化问题的初值,在此基础上调用SDP算法进行动态抓取力优化。以有摩擦点接触三指抓取长方体的仿真为例,该初值算法仅需100步迭代即可收敛,耗时0.471 ms,且初值与最优值的偏差小,在线计算量小。该方法可以用于多指机械手的抓取控制与仿真,能够实现对连续变化外力旋量的抓取力优化。In order to improve the efficiency of multi-fingered dynamic force optimization, a non- negative linear combination method for initial force caculation, combined with semidefinite program- ming（SDP） algorithm to optimize the dynamic contact force was proposed. According to the property that the set of grasp force is a convex cone, the external spinor was decentrolized as a linear combi- nation of 12 unit force vectors, and then the intial basic feasible solution for every unit force vector was caculated offline, the intial value of the contact force was then obtained by combination. Based on the above caculation, the SDP algorithm was called to optimize the dynamic contact force online. To take the three fingered point contact with friction for example, the non-negative linear combina- tion force optimization method can converged within 100 steps, which costs 0.471 ms, and the error between initial force and the optimization force is small, the online computation cost is also small. This method can be applied to nmlti-fingered robotic hand control and simulation, so that the object can be grasped in a force closure manner.

关 键 词：动态力优化 线性组合 多指抓取 抓取力初值 

分 类 号：TP242[自动化与计算机技术—检测技术与自动化装置]