The design of control software for robot swarms is a challenging endeavour as a swarm behaviour is the outcome of the entangled interplay between the dynamics of the individual robots and the interactions among them and with the environment. Automatic design techniques are a promising alternative to classic ad-hoc, code-and-fix design procedure and are especially suited to deal with the inherent complexity of swarm behaviours. Recently, information theory and complexity theory measures has been proposed for the analysis of single autonomous agent behaviour. Indeed, complex systems science may provide a corpus of theories and methods that enable the designer to formally and quantitatively analyse the dynamics of a robot swarm along with its internal information processes, and support the design of robot swarms. The long term goal is to use complexity measures to provide task agnostic merit factors for the automatic design procedures and to classify swarm tasks in terms of their intr insic complexity so as to optimally tune the complexity of individual robot software control systems as well as robot interactions. In this talk, Andrea Roli will illustrate the first results attained in applying complexity measures to the behaviour a group of robots and will outline the main future activities of this line of research. The work that will be presented has been done at IRIDIA, in collaboration with Mauro Birattari, during a visiting period Andrea Roli spent at ULB as a visiting professor within the ULB "Chaire internationale" programme 2016.
Swarm robotics, Complex systems, Complexity measures, Automatic design