Seminar 1: BrIAS Fellow Prof. Marco Nicotra
Systematic Design of Control Barrier Functions using Dynamic Safety Margins
Abstract: Control Barrier Functions (CBFs) have become a popular tool for constrained control due to their low computational cost and good performance. Unfortunately, the widespread use of CBFs is limited by the absence of a systematic method to design them for general systems. In this talk, we will draw a parallel between CBFs and Reference Governors (RGs). Doing so will allow us to use existing tools from the RG literature to systematically design CBFs for nonlinear systems with multiple state-and-input constraints and arbitrary relative degree.
Seminar 2: BrIAS Junior Fellow Dr. Elias Fernández Domingos
Delegation to AI in a collective Risk Dilemma
Abstract: The important contributions of Ostrom have identified a number of mechanisms that enable the correct management of local commons (e.g., community monitoring, punishment, institutions, voting). These mechanisms provide a social barrier to support sustainable decisions and prevent those that will have a future negative effect in society. Nevertheless, the spread of intelligent systems and artificial intelligence (AI) has affected significantly, not only the way humans acquire and share information, but also the way we make informed decisions regarding critical social questions such as climate action, sustainability, or compliance with health measures. In this talk, I will introduce the key factors that differentiate delegation to AI from delegation to other human beings, show experimental results on delegation to AI in a collective risk social dilemma, and highlight both the challenges and the potential opportunities that a hybrid human-AI society offers for solving important societal issues.
