Relative Localization in Colony Robots

Abstract

This project demonstrates the coordinated discovery and capture of a mobile target using a colony of low-cost robots. Many existing robot colonies use high-end components to guarantee results. We show that a colony of inexpensive robots, which alone are insufficient to the task, can together overcome their shortcomings. Information about the environment is gathered via simple one-dimensional sensors, such as IR beacons, and shared wirelessly. The robots localize relative to one another by sharing their 1D range data. All sensors are noisy and inaccurate, but when filtered, shared interpretation allows useful state information to emerge. No one robot is in charge or knows the entire state of the world, but as a communal intelligence the colony does. To search the world space for the target, the robots create regular formations and move as a coordinated mass. Once the heated target is located via pyroelectric sensors, the robots communicate the target's position and form a strategy to capture it using behavioral paradigms. We will demonstrate robots moving in a coordinated formation to surround and capture a target.