A single visual-servo controller of mobile robots with super-twisting control

• A single image-based controller that exploits the trifocal tensor is proposed.
• The robot reaches pose regulation using a single controller and without local minima.
• A virtual target avoids bad conditioning of measurements and control singularities.
• A super-twisting control scheme is introduced, which guarantees good robustness.
• The approach is valid for different types of central cameras.

This paper presents a novel approach for image-based visual servoing, extending the existing works that use the trifocal tensor (TT) as source for image measurements. In the proposed approach, singularities typically encountered in this kind of methods are avoided. A formulation of the TT-based control problem with a virtual target resulting from the vertical translation of the real target allows us to design a single controller, able to regulate the robot pose towards the desired configuration, without local minima. In this context, we introduce a super-twisting control scheme guaranteeing continuous control inputs, while exhibiting strong robustness properties. Our approach is valid for perspective cameras as well as catadioptric systems obeying the central camera model. All these contributions are supported by convincing numerical simulations and experiments under a popular dynamic robot simulator.