The Swarm-bot experience: Strength and mobility through physical cooperation

In this chapter, we study a collective robotic system that is inspired by the behavior of social insects. The system consists of mobile robots that can autonomously perceive and modify their environment. We investigate the problem of controlling a number of these robots in cooperation based tasks that are too difficult for the robots to solve when operating alone. In social insects, a range of sophisticated macroscopic phenomena can be observed at the level of groups, teams, and entire colonies [H ¨ olldobler and Wil-son (1990); Anderson and McShea (2001)].These phenomena are usually governed by self-organized processes. Such processes result from numerous local interactions between individual insects and between the individuals and their environment [Camazine et al. (2003)]. None of the individuals have any reference to a global template or pattern. Instead, each individual uses only local information to guide its actions. Designing collective robotic systems according to the principles of decentralized control and self-organization has a number of potential benefits. First of all, self-organized systems are inherently robust with respect to many types of failure. Typically, collective robotic systems have some inherent redundancy as they are comprised of many identical units. Thus, if some robots fail, other robots can take their place. Furthermore, as there is no centralized control in the form of, for instance , a leader robot, there is no single point of failure. Hardware-wise, the units of these robotic systems are typically relatively simple when compared to more traditional robotic systems consisting of a single monolithic robotic entity. Thus, collective robotic systems also tend to have fewer components inside each robotic unit that can fail. It is also typically true that the more units are involved in a given