The impact of catastrophic collisions and collision avoidance on a swarming behavior

Swarms of autonomous agents are useful in many applications due to their ability accomplish tasks a decentralized manner, making them more robust failures. Due the difficulty running experiments with large numbers hardware agents, researchers often make simplifying assumptions and remove constraints that might be present real swarm deployment. While away some is tolerable, we feel two particular have been overlooked: one, take up physical space, two, damaged collisions. Many existing works assume negligible size or pass through each other no added penalty. It seems possible ignore these using collision avoidance, but show an illustrative example this easier said than done. In particular, avoidance can interfere intended swarming behavior significant parameter tuning necessary ensure emerges as best while collisions avoided. We compare four different algorithms, which consider algorithms available. Despite putting effort into algorithm perform at its best, believe our results further research develop behaviors achieve goal avoiding non-negligible volume.