Locating RF Emitters with Large UAV Teams Paul Scerri School of Computer Science, Carnegie Mellon University, Pittsburgh, PA 15213, USA Robin Glinton School of Computer Science, Carnegie Mellon University, Pittsburgh, PA 15213, USA Sean Owens School of Computer Science, Carnegie Mellon University, Pittsburgh, PA 15213, USA Steven Okamoto School of Computer Science, Carnegie Mellon University, Pittsburgh, PA 15213, USA Katia Sycara School of Computer Science, Carnegie Mellon University, Pittsburgh, PA 15213, USA The rapidly improving availability of small, unmanned aerial vehicles (UAVs) and their ever reducing cost is leading to considerable interest in multi-UAV applications. However, while UAVs have become smaller and cheaper, there is a lack of sensors that are light, small and power efficient enough to be used on a small UAV yet are capable of taking useful measurements of objects often several hundred metres below them. Static or video cameras are one option, however image processing normally requires human input or at least computationally intensive offboard processing, restricting their applicability to very small UAV teams. In this paper, we look at how teams of UAVs can use very small Relative Signal Strength Indicator (RSSI) sensors whose only capability is to detect the approximate strength of a Radio Frequency (RF) signal, to search for and accurately locate such sources. RSSI sensors give at most an approximate range to an RF emitter and will be misleading when signals overlap. Applications of such UAV teams range from finding lost hikers or skiers carrying small RF beacons to military reconnaissance operations. Moreover, the core techniques have a wider applicability to a range of robotic teams that rely on highly uncertain sensors, e.g., search and rescue in disaster environments. pdflogo.jpg AAMAS07_0094_c561b354089ce24a543a444e368f5c21