双臂机器人神经网络解耦与路径规划算法研究  

Research of Dual-Arm Robot Neural Network Decoupling and Trajectory Planning Algorithm

作　　者：韦素云[1] 徐敏[2] 

机构地区：[1]包头钢铁职业技术学院,内蒙古包头014010 [2]南京航空航天大学机械结构力学及控制国家重点实验室,江苏南京211106

出　　处：《机械设计与制造》2017年第8期245-248,共4页Machinery Design & Manufacture

基　　金：国家自然科学基金(51105196)

摘　　要：针对双臂机器人的手臂运动控制问题,研究了其神经网解耦和路径规划算法。首先,对其双臂机械结构进行了分析,总结了各个关节对机器人末端位姿的影响;其次,通过MDH法建立了连杆坐标系,并给出了机械臂的运动学模型,进而依据手臂末端位置与姿态解耦的特点,通过三层BP神经网络与解析法相结合的方式研究了解耦逆运动学算法,避免了蚁群算法收敛较慢的问题;最后针对S曲线介绍了规划过程中各个参数的求解和适配方法,提出一种基于数值积分的B样条求解方法实现曲线拟合,优化空间曲线的平顺性;最后,对运动学算法与轨迹规划算法的有效性进行了验证。The neural network decoupling and trajectory planning algorithm are studied for motion control problem of dual-arm robot. Firstly, the mechanical structure of the arms is analyzed and the influence of each joint on the end pasition of the robot is summarized. Secondly, the connecting rod coordinate system is established by MDH method, and the kinematic model of the manipulator is given. The decoupling inverse kinematics algorithm is studied by means of three-layer BP neural network and analytic method, which avoids the problem that the ant colony algorithm converges slowly. Finally, the process of the simulation is introduced for the S-curve. The optimization of the spatial curve is carried out by using the B-spline solution method based on numerical integration. Finally, the validity of the motion algorithm and the trajectory planning algorithm is verified by the method of solving and fitting the parameters.

关 键 词：双臂机器人 运动学 轨迹规划 神经网络 

分 类 号：TH16[机械工程—机械制造及自动化] TP242.3[自动化与计算机技术—检测技术与自动化装置]