Gian Carlo Maffettone

giancarlo.maffettone[at]unina.it

PhD in: MERC
Ciclo: XXXVI

Project title: Controlling the collective dynamics of large-scale hybrid multi-agent systems

Supervisor(s): prof. Mario di Bernardo, prof. Maurizio Porfiri

  • Scarica il CV

  • Gian Carlo Maffettone is currently a Ph.D. student in Modeling and Engineering Risk and Complexity at Scuola Superiore Meridionale.

    He graduated cum laude in Automation Engineering in October 2020 at the University of Naples Federico II, with a research thesis on pattern formation in large scale multi-agent systems, under the supervision of Prof. Mario di Bernardo.

     

    Controlling the collective dynamics of large-scale hybrid multi-agent systems

    Tutors: prof. Mario di Bernardo and prof. Maurizio Porfiri.

    Multi-agent systems consists in ensembles of (nonlinear) dynamical systems interacting over a network with a certain (time-varying) topology.  When the number of involved individuals gets consistently large, we start talking about large-scale multi-agents systems. This scenario typically emerges in biology or social sciences where cell populations or social networks can include thousands or even millions of individuals or nanoparticles, and swarm robotics where the number of agents will become increasingly larger as technology to miniaturize devices and sensors improves.

    In this particular framework, to characterize and capture the dynamics of very large ensembles, continous mathematical descriptions (i.e. mean-field theory for example) are necessary. A crucial open problem is how to embed a local and distributed control strategy in such models, to steer the dynamics of the group towards a desired behavior. To this aim, hybrid mathematical models naturally arise, coupling discrete descriptions (ODEs for example) for the control part and continuous descriptions (PDEs for example) for the uncontrolled one. Notice that similar problems arise when dealing with small groups of dynamical systems, whose interaction is mediated by a fluid, as in schools of fishes.

    We want to develop tools to analyse and control such hybrid multi-agent systems, described by a combination of PDEs and ODEs toward the synthesis of feedback strategies to unfold and control their behavior. These tools will be then applied to the problem of investigating the emergence of social behaviour in zebrafish, a popular species for animal studies, and the design of robotic fish replicas that could be inserted in a zebrafish school to influence collective behaviour.