In this paper, a family of hybrid control algorithms is presented; where it is merged a free camera-calibration image-based control scheme and a direct force controller, both with the same priority level. The aim of this generalised hybrid controller is to regulate the robot-environment interaction into a two-dimensional task-space. The design of the proposed control structure takes into account most of the dynamic effects present in robot manipulators whose inputs are torque signals. As examples of this generalised structure of hybrid force/vision controllers, a linear proportional-derivative structure and a nonlinear proportional-derivative one (based on the hyperbolic tangent function) are presented. The corresponding stability analysis, using Lyapunov's direct method and invariance theory, is performed to proof the asymptotic stability of the equilibrium vector of the closed-loop system. Experimental tests of the control scheme are presented and a suitable performance is observed in all the cases. Unlike most of the previously presented hybrid schemes, the control structure proposed herein achieves soft contact forces without overshoots, fast convergence of force and position error signals, robustness of the controller in the face of some uncertainties (such as camera rotation), and safe operation of the robot actuators when saturating functions (non-linear case) are used in the mathematical structure. This is one of the first works to propose a generalized structure of hybrid force/vision control that includes a closed loop stability analysis for torque-driven robot manipulators.
This paper presents a design tool of impedance controllers for robot manipulators, based on the formulation of Lyapunov functions. The proposed control approach addresses two cha\-llen\-ges: the regulation of the interaction forces, ensured by the impedance error converging to zero, while preserving a suitable path tracking despite constraints imposed by the environment. The asymptotic stability of an equilibrium point of the system, composed by full non\-li\-near robot dynamics and the impedance control, is demonstrated according to Lyapunov's direct method. The system's performance was tested through the real-time experimental implementation of an interaction task involving a two degree-of-freedom, direct-drive robot.