Cannibalism, altruism and trophallaxis strategies among self-sustainable swarm robots

Energy sourcing and usage is a critical component in environmental swarm robotics. Populations of autonomous agents must gather energy from the environment and decide how to distribute it amongst themselves. Determining the optimum strategy for energy management across the swarm, with respect to the high-level goal of the population, remains a challenge. In this paper we explore three bio-inspired energy transfer strategies for self-sustainable swarm robots: Trophallaxis, Altruism and Cannibalism and build a simulation to evaluate the optimal strategy. Decentralised robot agents traverse a bounded environment and undertake terrain detection and food exploration tasks and the total rating of each simulation is recorded as a measure of mission success. Statistical results indicate that dynamic energy transfer can affect the performance of swarm robots significantly, with cannibalism and altruism being suitable for terrain coverage and trophallaxis being best for urgent tasks. This work shows the importance of implementing energy sharing strategies for a wide range of swarm applications, and the suggests that the optimal collaboration strategy is heavily influenced by the specific task goal.