A Multi-robot Chemical Source Localization Strategy Based on Fluid Physics: Theoretical Principles

Abstract

This book presents many examples of a primary advantage of physicomimetics—that its algorithms have provable properties of practical interest. This chapter presents yet another such example. In particular, by tackling the task of “chemical source localization” (i.e., using robots to locate the source of a toxic emission) from physics principles, we have invented an algorithm that is both provably correct and experimentally superior to its leading competitors for this task. The particular form of physics on which this algorithm is based is fluid dynamics. Robot movement is driven by a resultant force vector that propels a cohesive robot collective around obstacles and false alarms (e.g., regions of high chemical concentration that are not the source emitter) and toward the true emitter. Our algorithm is also the first to have a theoretically founded, rather than purely heuristic, emitter identification test. This book chapter focuses on analyses of our algorithm that not only provide behavioral guarantees to reassure the practitioner, but also provide deeper insights into why our algorithm is so successful.