A Neural Controller for Image-Based Visual Servoing of Manipulators With Physical Constraints

Area of Science:

• Robotics
• Computer Vision
• Control Systems

Background:

• Visual servoing of manipulators faces challenges with feature error convergence and joint limit safety.
• Existing methods often require complex matrix computations like pseudoinversion.

Purpose of the Study:

• To propose an image-based visual servoing scheme for eye-in-hand manipulators.
• To ensure rapid convergence of feature errors and compliance with joint limits.

Main Methods:

• An image-based visual servoing scheme is developed for eye-in-hand configurations.
• The method avoids pseudoinversion of the image Jacobian and inversion of the forward kinematics Jacobian.

Main Results:

• The proposed scheme guarantees asymptotic convergence of feature errors to zero.
• It ensures compliance with manipulator joint angle and velocity limits.
• Simulations on a PUMA560 manipulator validate the scheme's effectiveness.

Conclusions:

• The novel visual servoing scheme effectively addresses key challenges in manipulator control.
• It offers a computationally efficient and safer alternative to existing methods.