微创外科手术机器人运动学分析与研究  

Kinematics Analysis of Minimally Invasive Surgery Robot

作　　者：何志锋[1] 朱坚民[1] 宋成利[2] 徐兆红[2] 

机构地区：[1]上海理工大学机械工程学院,上海200093 [2]上海理工大学医疗器械与食品学院,上海200093

出　　处：《计算机仿真》2014年第4期412-415,439,共5页Computer Simulation

基　　金：上海市科学技术委员会基金资助项目(10PJ1407800);上海高校选拔培养优秀青年教师科研专项基金资助项目(slg10019)

摘　　要：在微创手术机器人的运动学问题的研究中,微创外科手术机器人与传统工业机器人在精度上要求不同,后者强调的是重复精度,而前者则要求有良好的绝对位置精度,而且手术过程中要求有良好的实时性能,传统方法求解机器人逆运动学方程运算量大,速度慢,不利于实现实时控制。为解决上述问题,采用kvenberg-Marquardt BP神经网络算法,建立了机器人末端执行器位姿与各关节角之间的映射关系,进而实现了机器人逆运动学快速准确求解;在Matlab平台上建立2-DOF机器人逆运动学神经网络模型,并进行程训练和仿真,结果表明,BP神经网络学习精度高,收敛速度快,可满足手术机器人精确求解逆运动学的要求。Based on the minimally invasive surgery robot made by ourselves, this paper analyzed a 6 DOF multi joint robot in kinematics. And we focused on the mothod of solving inverse kinematics. Minimally invasive surgery robot is different from the traditional industrial robots on the precision requirement. The latter emphasizes the repeat precision, while the former requires a well absolute positioning accuracy. And it requires a good real time perform ance in the process of operation. This paper used the Levenberg Marquardt BP neural network algorithm to establish the robot end effector pose with the mapping between the angle of joint, thus achieving a fast and accurate robot in verse kinematics solution. At last, through the simulation of two link robotic manipulator, we proved the method is feasible.

关 键 词：微创手术机器人 神经网络 逆运动学 

分 类 号：TP242[自动化与计算机技术—检测技术与自动化装置]