Task allocation in robot swarms for time-constrained tasks

Robotic systems offer auspicious solutions for many real life applications, where robots can replace humans in many dangerous and critical tasks. A promising branch of robotic systems is swarm robotics. This branch has stimulated a significant interest by researchers belonging to different fields such as biology, mathematics, computer science and robotics. A swarm robotics system consists of a large number of simple robots, which can vary between hundreds and thousands. These systems are able to execute complex tasks, which are beyond the capability of a single robot system. They offer a range of advantages including: fault tolerance, flexibility and scalability. The high level of fault tolerance offered by swarm robotics is due to their large size, as functioning robots can substitute malfunctioning or even broken robots. Swarm robotics is a flexible system which is evinced by its ability to be involved in a large spectrum of applications without the need to modify the robots’ hardware. Furthermore, it is a scalable system, where changing the number of robots does not affect the algorithm of the solution. Tasks which require massive parallelism, redundancy or to be executed in complex environments, are good examples where swarm robotics can provide appropriate solutions. This thesis focuses on a special category of tasks, referred to as timeconstrained tasks. In time-constrained tasks a specific amount of work should be accomplished on the task within a given deadline. Since tasks with time constraints represent a large category of the real-world applications, swarm robotics will have the need to participate often on executing such tasks. Our work concentrates on developing task allocation strategies which are supposed to assign robots swarms to time-constrained tasks under the condition of the task deadlines. Designing swarm robotics systems can be performed following a bottomup paradigm, which is referred to as a microscopic-macroscopic design. It starts from designing the single robot behavior and aims to achieve a desired global behavior. Another paradigm which can be used in designing swarm robotics systems is the top-down one, which is referred to as the macroscopic-microscopic design. It starts from designing the swarm behavior and ends with designing the behavior of the single robots. Selecting between the two paradigms depends on the goal of the study in addition to the focus of the swarm application. In this thesis, global constraints, namely the task deadlines, are required to be fulfilled. Therefore, designing the system starts from the global level and aims to achieve individual robots design that satisfies the global requirements. Hence, the design paradigm used in this thesis, is the macroscopic-microscopic one.