论文部分内容阅读
提出了一种对机器人机械臂各关节的动力学参数进行离线识别的新方法。该方法在实现途径上简单有效,无需附加传感器。通过控制机械臂各待辨识关节按照周期性傅里叶级数轨迹运动,读取关节电流以获取关节力矩值便可实现对机械臂连杆动力学参数进行识别。该识别算法通过将待识别参数分为线性部分和非线性部分,并对其分别采用线性最小二乘法和粒子群算法进行解算。实现过程简单,不仅能识别机械臂的动力学参数,还能识别动摩擦系数和黏滞摩擦系数。通过在新松七自由度协作机器人上进行实验,验证了该方法的可行性和有效性。
A new method for offline recognition of dynamic parameters of robot manipulator joints is proposed. This method is simple and effective way to achieve, without additional sensors. By controlling the robot arm to be identified joints in accordance with the periodic Fourier series trajectory to read the joint current to obtain joint torque value can be achieved on the robot arm kinematics parameters identification. The recognition algorithm divides the parameters to be identified into linear part and non-linear part, and solves it by linear least squares method and particle swarm optimization respectively. The realization of the process is simple, not only to identify the kinematic parameters of the manipulator, but also to identify dynamic friction coefficient and viscous friction coefficient. Experiments on the Xinsong seven-DOF collaborative robot show that the method is feasible and effective.