| Summary: | Formations are the spatial organization of objects or entities according to some predefined pattern. They can be found in nature, in social animals such as fish schools, and insect colonies, where the spontaneous organization into emergent structures takes place. Formations have a multitude of applications such as in military and law enforcement scenarios, where they are used to increase operational performance. The concept is even present in collective sports modalities such as football, which use formations as a strategy to increase teams efficiency. Swarm robotics is an approach for the study of multi-robot systems composed of a large number of simple units, inspired in self-organization in animal societies. These have the potential to conduct tasks too demanding for a single robot operating alone. When applied to the coordination of such type of systems, formations allow for a coordinated motion and enable SRS to increase their sensing efficiency as a whole. In this dissertation, we present a virtual structure formation control synthesis for a multi-robot system. Control is synthesized through the use of evolutionary robotics, from where the desired collective behavior emerges, while displaying key-features such as fault tolerance and robustness. Initial experiments on formation control synthesis were conducted in simulation environment. We later developed an inexpensive aquatic robotic platform in order to conduct experiments in real world conditions. Our results demonstrated that it is possible to synthesize formation control for a multi-robot system making use of evolutionary robotics. The developed robotic platform was used in several scientific studies.
|
|---|