Modeling and Prediction of Emergent Behaviors in Distributed Robotic Systems
Ref. Prof. Armando Tacchella
In the last decade, interest in distributed systems of mobile robots has been increasing. The distributed approach foresees teams of robots - also called swarms - that are able to perform cooperative tasks without any centralized coordination, and it allows for robust, flexible, efficient and scalable solutions. However, in spite of the simple algorithms governing each individual, the swarm may present so called emergent behaviors, i.e., regular, complex, not explicitly programmed collective dynamics. Compared to the amount of research on the design of robotic teams, little has been done to investigate the issue of emergent behaviors, and this leaves important open questions as far as safety and efficiency of the swarm are concerned. The aim of this thesis is to provide a methodology whereby formal models of robotic swarms can be analyzed automatically using computer aided verification and reasoning. The ultimate goal is to provide a precise, yet practically feasible way to predict and bound emergent behaviors in such systems without performing extensive simulations or expensive on-line testing.
Distributed Robotic Systems, Formal Models, Computer-aided Verification and Reasoning
Excellent programming skills, good knowledge of structured specification methods (e.g., UML), basic software engineering practices, including requirement specification and functional testing; some knowledge of formal methods and/or logic-based reasoning is a plus.
Project start date: from January 2016
Collaborations: Istituto Italiano di Tecnologia (Italian Institute of Technology) – iCub facility