Title of Presentation: Controlled Reconfiguration of Robotic Mobile Sensor Networks using Distributed Allocation Formalisms
Primary (Corresponding) Author: Antidio Viguria
Organization of Primary Author: Georgia Institute of Technology
Co-authors: Ayanna Howard
Abstract: In this paper, we discuss the realization of a robotic mobile sensor network that allows for controlled reconfiguration of sensor assets in a decentralized manner. The motivation is to allow the construction of a new system of in-situ science observations that requires higher spatial and temporal resolution models that are needed for expanding our understanding of Earth system change. These observations could enable recording of spatial and temporal variations in environmental parameters required for such activities as monitoring of seismic activity, monitoring of civil infrastructures, and detection of toxic agents throughout a region. In most sensor network applications, individual sensor agents collect information about their environment and neighbors using peer-to-peer communication. Unfortunately, as the size of the network increases, bandwidth limitations and the absence of feasible communication channels severely limits the possibility of conveying and using global information. Thus, it cannot be assumed that each sensor agent has complete information about the states of every other agent in the network. And yet, a network formation is inherently a global property and, as such, novel solutions must be implemented for the reconfiguration process to successfully occur. To enable controlled reconfiguration of sensor assets, we need a formal method that allows the multi-agent sensor network to be redeployed and reprogrammed with relative ease. One formalism, called Graph Grammers, allows a compact representation of graph interactions. By extending this formalism, we can represent our robotic mobile sensor network using connected graphs in which vertices of the graphs represent the sensor agents, and connected edges are established when agents are within communication range of each other. This formalism, coupled with local control laws, allows robust control of the network to achieve global tasks in natural environments. We discuss this extended formalism in detail and the results of its applications to a science-driven coverage task.