关键词: 机器人, SPSA, Kalman滤波, 无标定视觉伺服, 图像雅克比

Abstract: Considering the problem of robot uncalibrated visual servoing, this paper presents a method for online estimation of image Jacobian matrix based on Kalman filter optimized by simultaneous perturbation stochastic approximation algorithm. This method takes the robot image Jacobian matrix as the system state, and uses Kalman filter to observe the system state. In order to improve the performance of the filter, the simultaneous perturbation stochastic approximation algorithm is used to optimize the filter parameters. This method is used to estimate the image Jacobian matrix and to design the control strategy, which can avoid complicated system calibration process. The simulation results indicate that the proposed method can achieve the visual positioning of the 6-degree of freedom robot with high accuracy and stability under the uncalibrated situation.

Key words: robot, SPSA, Kalman filter, uncalibrated visual servoing, image Jacobian