Distributed control strategies for cooperative and collaborative object transportation via Multi-Robot systems

This research deals with the development of distributed control strategies for object transportation within multi-robot frameworks, encompassing both fully autonomous cooperative manipulators and collaborative human-robot interaction. The first part of the work investigates the cooperative transportation problem, where a team of robots is tasked with manipulating a shared load relying exclusively on local peer-to-peer communication. To address the scalability and robustness limitations inherent in centralized architectures, a novel methodology for the distributed estimation of desired trajectories and exchanged forces is introduced. Firstly, an appropriate kinematic formulation is established to describe coordinated motion tasks in terms of absolute and relative variables. Within this framework, consensus-based estimators enable each agent to compute the estimates of the desired trajectories, which are subsequently exploited for real-time end-effector reference calculations. Secondly, distributed control schemes are developed to regulate the internal forces exerted on the object while limiting external wrenches arising from the interaction with the environment via an admittance control framework. The effectiveness of the overall architecture is validated through both high-fidelity simulations and experimental campaigns on homogeneous and heterogeneous manipulator setups, demonstrating a significant reduction in internal stress on the transported object and robust handling of external environmental contacts. The convergence of the estimators and the stability of the distributed control schemes are rigorously assessed through Lyapunov-based analyses. The second part of the work investigates Human-Robot Collaboration (HRC) by integrating the operator as an agent-in-the-loop through an advanced Augmented Reality (AR) framework. Leveraging the developed AR features, a bidirectional communication flow is established to address dynamic role allocation and the exchange of intentions during shared transportation tasks. The proposed methodology enhances the operator's perception by providing real-time feedback on the forces exerted by the robot, while allowing for intuitive feed-forward trajectory planning. A key contribution of this research is the implementation of a flexible autonomy scheme, enabling the operator to dynamically switch between a leader-follower configuration and a shared-trajectory mode. In the latter, a distributed admittance control strategy ensures the minimization of interaction forces and guarantees operational safety. Furthermore, the framework incorporates safety-critical monitoring of both human and robotic signals, reducing the physical workload for the operator. Experimental validations demonstrate that while AR-based interaction might initially increase cognitive load for novice users, a properly tuned level of robot autonomy ensures a seamless and effective collaborative experience.