Dynamic Jacobian Identification Based on State-Space for Robot Manipulation

Abstract

This paper focused on the dynamic Jcobian identification for robot visual feedback control in unstructured environments, there is the robotic manipulator without kinematics model, and the camera without calibration. First, a visual servoing scheme based on Kalman filtering (KF) to model-independent robotic manipulation is proposed. The Jacobian identification were solved by formatted as state estimation problems, the desired Jacobian is optimal estimated under minimum-variance condition to achieves robotic precise converge to the desired pose. Our method, without requiring modeling parameters, thus this avoids the corrupted performances cause by calibration errors. To demonstrate the performance of proposed scheme, the experimental results have been presented using eye-in-hand configurations.