环境约束可重构机械臂模块化力/位置控制

针对受环境约束的可重构机械臂系统,提出了一种自适应神经网络模块化力/位置控制方法.利用雅克比矩阵将机械臂末端与环境接触力映射到各关节,将系统动力学模型描述成一组通过耦合力矩相关联的子系统集合,通过控制各子系统的位置和力矩来达到控制末端执行器位置和接触力的目的.利用神经网络估计可重构机械臂系统的非线性项和交联项,通过自适应更新律在线估计神经网络权值函数,并引入滑模控制项补偿估计误差,从而保证闭环系统渐近稳定.最后,在不改变控制器参数的条件下对2个不同构形的2自由度可重构机械臂进行数值仿真,结果验证了所设计控制器的有效性.

Modular Position/Force Control for Environmental Constrained Reconfigurable Manipulator

This paper proposed an adaptive neural network modular position/force control method for environmental constrained reconfigurable manipulator. Mapping the end point force of the reconfigurable manipulator to each joint module, describe the entire dynamic model by a set of interconnected subsystems by coupling torques, and the position and contact force of the end point should be obtained by controlling the angle and torque of each subsystem. By virtue of the neural networks, the nonlinear and interconnected items are estimated, and the weights of these neural networks can be adaptively updated on line, meanwhile, the estimated errors are removed by a sliding mode control item, and the stabiltiy of the closed loop system should be granteed. Finally, the numerical simulation on two different 2 DOF reconfigurable manipulators shows the effectiveness of the proposed method without changing control parameters.