Special GRASP Seminar – Stephanie Gil, Massachusetts Institute of Technology, “Optimizing Communication in Air-Ground Robot Networks using Decentralized Control”

Abstract: This talk concerns the development of a distributed controller that positions a team of aerial vehicles in a configuration to optimize communication-link quality, providing support to a team of ground vehicles performing a collaborative task. We propose a gradient based control approach where agents’ positions locally minimize a physically motivated cost function. The talk will detail the formulation of a cost function that incorporates a continuous, physical model of signal quality, SIR and the development of a non-smooth gradient-based controller that positions aerial vehicles to achieve optimized signal quality amongst all vehicles in the system. Details of the convergence proof for this controller, specifically how we are able to incorporate non-differentiability due to agents moving in or out of communication with one another, will also be presented. Lastly, we will derive conditions that guarantee the aerial vehicles will never disconnect the connectivity graph. The talk will conclude with aggregate results for our controllers’ performance on a group of AscTec Hummingbird quadrotors.