Area coverage in two-dimensional grid worlds using computation-free agents

This work proposes a novel solution to the problem of covering a bounded grid world using a swarm of robotic agents. The controller requires no run-time memory and only few, discrete sensory inputs. Two variants of the solution to the problem are studied, one effectively modulating the sensing range based on the agent’s context. It is found that during the dispersion, the controller with sensing range modulation outperforms the default controller in terms of speed and evenness of the dispersion. Due to its simplicity, the solution could be realised on swarms of agents with ultra-low power and computational requirements, making it potentially relevant for large-scale swarm applications.