Distributed Data Storage and Fusion for Collective Perception in Resource-Limited Mobile Robot Swarms

In this letter, we propose an approach to the distributed storage and fusion of data for collective perception in resource-limited robot swarms. We demonstrate our a semantic classification scenario. consider team mobile robots, which each runs pre-trained classifier known accuracy annotate objects environment. provide two main contributions: (i) decentralized, shared structure efficient retrieval annotations, specifically designed low-resource robots; (ii) voting-based, decentralized algorithm reduce variance calculated annotations presence imperfect classification. discuss theory implementation both contributions, perform extensive set realistic simulated experiments evaluate performance approach.