Ants in Parking Lots

Ants provide an attractive metaphor for robots that “cooperate” in performing complex tasks. What, however, are the algorithmic consequences of following this metaphor? This paper is a step toward understanding the algorithmic strengths and weaknesses of ant-based computation models. We study the ability of ant-robots that are essentially mobile finite-state machines to perform a simple path-planning task called parking, within fixed, geographically constrained environments (“laboratory floors”). This task: (1) has each ant head for the nearest corner of the floor and (2) has all ants within a corner organize into a maximally compact formation. Even without using (digital analogues of) pheromones, many initial configurations of ants can park. These configurations include: (a) a single ant that starts anywhere along an edge of the floor and (b) any assemblage of ants that begins with at least two ants adjacent to one another. In contrast, a single ant on a one-dimensional “floor” cannot park, even with the help of (volatile digital) pheromones.