Robots visual servo control with features constraint employing Kalman-neural-network filtering scheme

• A NNAKF scheme is constructed for Robot manipulation in uncalibrated environment.
• The Jacobian identification problems were solved with state estimation techniques.
• The neural network served as an error estimator for compensation the errors of KF.
• The proposed IBVS approach with robust stability under features kept within FOV.

This paper presents an image-based servo control approach with a Kalman-neural-network filtering scheme for robots manipulation in uncalibrated environment. The image Jacobian on-line identification problems are firstly addressed by introducing the state estimation techniques, which have been incorporated neural network assists Kalman filtering (NNAKF). In fact, this is, the neural network (NN) can serve to play exactly the role of the error estimator, has the task of compensate the errors of Kalman filtering (KF). Then, by employing the NNAKF scheme, the proposed image-based servo control approach has guaranteed the robustness with respect to destabilized system attached dynamic noises, as well as the image features are constrained in field-of-view (FOV) of the camera. Furthermore, it is without requiring the intrinsic and extrinsic parameters of the camera during visual servoing tasks. To demonstrate further the validity and practicality of proposed approach, various simulation and experimental results have been presented using a six-degree-of-freedom robotic manipulator with eye-in-hand configurations.