Building a Decentralized Active Sensor Network

Abstract

This paper describes the theory and implementation details of an indoor Active Sensor Network (ASN). An ASN is comprised of a number of sensing platforms connected into a network. Some or all of the platforms are controllable and, in that sense, are active. Compared to related work, the approach described here strictly adheres to scalable decentralized algorithms for both the data fusion and the decision-making layers of the system. On the data fusion level, the information filter is used to combine observations of point features. On the control level, a decentralized planning algorithm allows individual nodes to coordinate their activities autonomously. Parts of the architecture have been verified on a network of UAVs and a team of indoor robots.