Autonomous Area Search Using Market-Based Assignment in Multi-Vehicle Unmanned Aerial Systems

Multi-vehicle unmanned aerial systems (UASs) are becoming more capable than ever before, making them increasingly suitable for complex tasks and behaviors. With increasingly sophisticated inter-vehicle coordination, human control can be replaced by human supervision of these systems' autonomously developed courses of action. Market-based auction algorithms provide one distributed mechanism that can be used to plan complex tasks such as area search by autonomously decomposing larger tasks and assigning subtasks to individual agents. This paper describes an auction algorithm implementation for planning and control of an area search by a UAS of fixed-wing and quadrotor unmanned aerial vehicles (UAVs). We test our implementation in three different search areas with system sizes ranging between two and 24 UAVs. We compare our results against the G-Prim algorithm and an idealized or "perfect" search. Results from both simulation and live-flight testing are discussed.