This paper considers the cooperative manipulation of a cable-suspended load with two generic aerial robots without the need of explicit communication. The role of the internal force for the asymptotic stability of the beam position-andattitude equilibria is analyzed in depth. Using a nonlinear Lyapunov-based approach, we prove that if a non-zero internal force is chosen, then the asymptotic stabilization of any desired beam attitude can be achieved with a decentralized and communication-less master-slave admittance controller. If, conversely, a zero internal force is chosen, as done in the majority of the state-of-the-art algorithms, the attitude of the beam is not controllable without communication. Furthermore, we formally proof the output-strictly passivity of the system with respect to an energy-like storage function and a certain input-output pair. This proves the stability and the robustness of the method during motion and in non-ideal conditions. The theoretical findings are validated through extensive simulations.

Cooperative manipulation is a basic skill in groups of humans, animals, and in many robotic applications. Besides being an interesting challenge, communication-less approaches have been applied to groups of robots in order to achieve higher scalability and simpler hardware and software design. We present a generic model and control law for robots cooperatively manipulating an object, for both ground and floating systems. The control method exploits a leader-follower scheme and is based only on implicit communication (i.e., the sensing of contact forces). The control objective mainly consists of steering the object manipulated by the swarm of robots to a desired position and orientation in a cooperative way. For a system with just one leader, we present analytical results on the equilibrium configurations and their stability that are then validated by numerical simulations. The role of object internal forces (induced by the robots through contact forces) is discussed in terms of convergence of the object position and orientation to the desired values. We also present a discussion on additional properties of the controlled system that were investigated using thorough numerical analysis, namely, the robustness of the system when the object is subject to external disturbances in non-ideal conditions, and how the number of leaders in the swarm can affect the aforementioned convergence and robustness.

No abstract