Receding Horizon Synthesis and Dynamic Allocation of UAVs to Fight Fires

Abstract

The need for more robust and trustworthy systems to fight wildfires stems from an annual economic burden exceeding $63.5 billion within the United States, elaborated in [1]. Current uses of unmanned aerial vehicles (UAVs) in such a pursuit typically provide “eyes in the sky”, and these vehicles may one day be capable of fighting such fires on their own, as observed in the small-scale test case of [2]. From such, a fleet of automated UAVs could potentially combat wildfires faster and more efficiently than a team made of only human operators while greatly reducing the danger to human life and property. Furthermore, such an approach could help increase public trust in advanced robotics in ways that directly impact people’s lives. Creating a system to achieve this task requires advancements in both the physical hardware and the AI software to control such a fleet. Our work explores the latter through the use of high-level controllers formed by formal methods, specifically reactive synthesis.