LAUSR

The formation of collaborative robotic teams for task execution often requires coordination in both space and time, with robots gathering in close vicinity and concurrently executing the task. For decentralised robotic swarms constituted of minimalist agents unable to communicate and plan ahead, a probabilistic approach might ensure that tasks are executed at the maximum possible rate by means of opportunistic team formation. We consider here the case of strictly-collaborative tasks of two typeseasy and hardeach requiring a specific number of agents to concurrently work in the same area. We show how task execution can be improved by adaptive behavioural strategies that (i) change the random movements of robots to bias their distribution towards areas where hard tasks are present, and (ii) specialise the robots behaviour to facilitate the formation of teams tailored to the one or the other task type. Experiments with simulated and real swarms of Kilobotsdemonstrate the suitability of the proposed approach, opening to future applications in micro/nano-robotics.