Hybrid Decentralized Control System for Communication Aware Mobile Robotic Teams

Enabling a team of robots to self-organize into a multi-hop ad hoc network as it simultaneously completes a task requires a system architecture that controls both the motion of each robot and their communication variables. In this paper, we consider this objective and propose a hybrid architecture composed of both centralized and decentralized components. This novel architecture utilizes the strengths of each component while limiting the drawbacks. The resulting system is therefore able to operate in more complex environments than decentralized systems, requiring less coordination overhead than centralized systems. We demonstrate through simulation that our hybrid system has the ability to successfully complete a task while navigating complex environments, simultaneously avoiding local minima commonly encountered in decentralized systems. Furthermore, we demonstrate through experimentation the ability of our hybrid system to achieve equal, if not greater, end-to-end date rates in comparison to centralized systems with lower coordination overhead.