Naslov
Long-term pattern formation and maintenance for battery-powered robots

Autori
Li, Guannan ; Švogor, Ivan ; Beltrame, Giovanni

Izvornik
Swarm Intelligence (1935-3812) 13 (2019), 1; 21-57

Vrsta, podvrsta i kategorija rada
Radovi u časopisima, članak, znanstveni

Ključne riječi
Long-term autonomy ; Energy-aware ; Multi-robot teams ; Graph ; Battery

Sažetak
This paper presents a distributed, energy-aware method for the autonomous deployment and maintenance of battery-powered robots within a known or unknown region in 2D space. Our approach does not rely on a global positioning system and therefore allows for applications in GPS-denied environments such as underwater sensing or underground monitoring. After covering a region, our system maintains a formation and uses an arbitrary number of charging stations to prevent robots from fully discharging. Analyzing the topology of the network formed during robot deployment, we generate virtual recharging trees which the robots use to navigate toward a nearby charging station when needed. All robots that leave the formation are replaced by their neighbors, maximizing the effective coverage provided by the system. We demonstrate the capability of our methods using models, a physics-based simulator, and experiments with real robots.

Izvorni jezik
Engleski

Znanstvena područja
Računarstvo, Informacijske i komunikacijske znanosti

POVEZANOST RADA