Robot manipulator visual servoing based on image moments and improved firefly optimization algorithm-based extreme learning machine.

We propose an improved extreme learning machine (ELM) to solve the decoupling problem between the camera coordinates and the image moment features for robot manipulator image-based visual servoing system, that is, determine the nonlinear relationship between them. First, an improved firefly optimization algorithm (IFOA) based on an adaptive inertial weight and individual variations is proposed. Then, the IFOA is optimized the weight and hidden bias in ELM algorithm; this improves the training accuracy of the ELM. Finally, the improved firefly optimization algorithm is integrated into ELM (IFOA-ELM) to solve the decoupling problem and ensure stable performance. The results of experiment show that the estimated error of the rotation angle around the camera frame in the visual servoing system determined by the IFOA-ELM algorithm is less than 0.25°, confirming that the proposed algorithm exhibits good robustness and stability. • A novel extreme learning machine (ELM) is used to solve the decoupling problem of the nonlinear mapping in visual servoing. • A new firefly algorithm uses an adaptive inertial weight and individual variations for optimization and exhibits faster convergence. • The input weight, bias, and output weight of the hidden layers of the ELM are optimized by the improved firefly optimization algorithm. [ABSTRACT FROM AUTHOR]