假手多指抓取的动态力优化仿真

Simulation for dynamic grasping force optimization of multi-fingered robotic hand

为完善假手康复训练系统,构建基于虚拟仪器的信号采集及多指抓取控制平台。该平台采用虚拟现实建模语言(virtual reality modeling language,VRML)组建分布式虚拟环境,基于多通道肌电信号,采用支持向量机预测抓取模式,通过骨骼算法确定复杂曲面接触条件下的接触点及法线方向。采用非负线性组合算法求取接触力初值,结合梯度流算法实现抓取力的动态优化,通过仿真实验验证其可行性。结果表明,骨骼算法用于抓取点计算,结果准确,动态力优化算法计算效率较高,且易于集成。

In order to overcome the imperfects of rehabilitation systems,a signal collection and multi-finger grasping control platform was built by virtual instrument modelling techniques. This platform was a distributed virtual reality environment which was constructed through VRML. The multi-channel electromyography signals were collected first,and then the support vector machine was utilized to predict the grasp pattern; afterwards,the contact points and the contact normal were decided by skeleton algorithm; at last,the dynamic grasp force was optimized by the gradient flow algorithm and the initial value was calculated by non-negative combination methods. Simulation was conducted to evaluate the feasibility of the system. The results show that the skeleton algorithm can accurately calculate the grasp points and efficiently optimize the force,both the two algorithms can be integrated and have a fairly good practical value.