鳍条效应软体采摘机械手建模与试验

Modeling and experiment of soft robotic picking hand based on fin ray effect

针对现有鳍条效应软体手指建模分析方法难以兼顾高精度与小运算量的问题,本研究以课题组设计的苹果采摘软体机械手为对象,在观察软体手指变形的基础上,提出将鳍条效应软体手指等效为串连铰链四杆机构;运用虚功原理推导了给定变形状态下软体手指各处接触力与驱动力矩计算方法;提出用线弹性扭簧模型描述变形恢复力矩,利用改进的粒子群算法求解各级四杆机构扭簧劲度系数;基于有限元模型,结合二次开发建立了给定关节转角和充气压力下驱动力矩的BP神经网络静态模型。搭建试验平台对3个不同大小的仿真苹果在3个不同高度位置下进行抓取力测量试验。试验结果表明,所建立的力学模型计算抓取力相对误差绝对值小于8.6%,与有限元模型精度相当;驱动力矩测量试验结果表明关节神经网络模型计算值与测量值变化趋势相同,相对误差小于12.7%。

Aiming at the problem that it is difficult for the existing modeling and analysis methods of soft finger based on fin ray effect to meet the requirements of high precision and small amounts of calculation,we proposed approximately regarding the soft finger as concatenated four-bar mechanisms by observing the soft finger deformation.The theoretical model to solve contact force and actuating torque was established based on the principle of virtual work when the deformation state is given.The deformation recovery torque was modeled using linearly elastic torsional spring,and then an improved particle swarm algorithm was programmed to solve the stiffness coefficient of the torsional spring of each phalanx.Based on the Abaqus finite element model and combined with the secondary development,the relationship between actuating torque and pressure,joint angle was obtained,and then fitted with the aid of BP neural network.Last,an experiment platform was built to measure the grip force of the soft robotic picking hand.The results showed that the absolute relative error of the model was less than 8.6%,which was equivalent to that of the finite element model.The values of actuating torque calculated by BP neural network shared the same trend with the values measured and the relative error was under 12.7%.