Broadcast of local eligibility for multi-target observation.pdf

Broadcast of Local Eligibility for Multi-Target Observation Barry Brian Werger and Maja J Mataric USC Robotics Research Lab, Los Angeles, CA 90089-0781 Abstract. While numerous researchers have investigated group behavior of robots which are each controlled in a behavior-based manner, none have yet thoroughly in- vestigated the possibilities of extending the port-arbitrated behavior (PAB) paradigm across networks of robots. We present an extension to the well-dened PAB tech- niques of behavioral interaction which provides standard abstractions for nessaging, inhibition, and suppression over IP networks. The Broadcast of Local Eligibility is a general technique built from these abstractions that allows fully-distributed, ex- ible team coordination. We present a BLE approach to the CMOMMT multi-target observation problem, implemented on a team of physical robots. 1 Introduction While numerous researchers have investigated group behavior of robots which are each controlled in a behavior-based manner, none have yet thoroughly investigated the possibilities of extending the port-arbitrated behavior [12] paradigm across networks of robots. While it has often been hypothesized that there need be no distinction between inter-robot and inter-behavior communication e.g., [2], to our knowledge no previous system has provided standard tools that allow port-based messaging, suppression, and inhibition between behaviors on separate networked robots. Our intention is to demonstrate that behavior-based systems restricted to well-dened port-arbitrated interactions can scale to higher levels of com- petence than is generally assumed. Specically, we show that when the port- arbitration paradigm is extended across networks, the resulting systems are able to dynamically recongure themselves in order to eciently allocate re- sources in response to task constraints, environmental conditions, and system resources. We introduce the Broadcast of Local Eligibility (BLE) as a general to