Data Driven Invariance Workshop

Abstract

Invariance is an important tool in ensuring the safety of autonomous and non-autonomous systems alike. However, determining the invariant set or control invariant set of a dynamical system remains a challenge -- either requiring synthesis of a model-specific scalar function or a method of calculating a system's one-step reachable set. With advances in machine learning, data driven methods present an interesting new direction for invariant set synthesis for known and unknown systems. However, many challenges remain – both in incorporating the uncertainty inherent to model-free analysis, and in dealing with the burdensome sampling requirements typical of data-driven methods.

This workshop brings together highlighted speakers who are tackling the challenges and opportunities of data driven invariance with a variety of techniques, including set based methods, barrier and Lyapunov function methods, and safety filters, to holistically survey the current state of the art. A rapid interactive/poster session will provide an opportunity for in-depth discussion and connection across the community.

Register for the workshop now! 

Speakers

Poster Session

We are excited to announce that our workshop will be holding a poster session for recent results in data driven invariance. A rapid interaction session will provide the opportunity to present results and interact with members of the community. 

To participate: Just email the organizers with basic details about your prospective poster and talk.

• Email: Leila.bridgeman@duke.edu

• Subject: CDC Workshop Poster Session

• Include: Talk/poster title, abstract, & your contact info