Scalability of robot swarms when applied to maze solving

Swarm robotics offers an intriguing approach to many real-world engineering problems. Such tasks that require covering a large search space or that involve working in potentially hazardous environments naturally lend themselves to robotic and/or swarm solutions. This research examines the effect of changing swarm size on the performance of several different maze-solving algorithms, including random movement, obstacle avoidance, and wall-following algorithms. The choice of algorithms was limited to ones that (a) do not require direct bot-to-bot communication, (b) use bots with a limited sensing range and (c) do not require the bots to know their position. Two different robots were used in the experimental trials: the HEXBUG Larva and the e-puck mobile robot. Trials were conducted with two, four, eight, and (with the Larva) sixteen bots. Results are presented for each of the trials as well as discussion on swarm performance as related to swarm size and bot intelligence. The results show that a small swarm that is more sophisticated or intelligent can perform as well as a larger swarm made up of less intelligent bots.