Supervisory Control for Collision Avoidance in Vehicular Networks using Discrete Event Abstractions
Add to Google Calendar
We will start with an introduction to supervisory control of discrete event systems (DES). We will discuss relevant system-theoretic properties for DES such as controllability, nonblockingness, and maximal permissiveness, that arise in the solution of supervisory control problems. Then we will present on-going research on the problem of collision avoidance at vehicular intersections for a set of controlled and uncontrolled vehicles that are linked by wireless communication. Each vehicle is modeled by a first-order system with bounded model uncertainty. We construct a finite DES abstraction of the underlying system that preserves the desired property of safety (i.e., no collisions). We formulate the problem of collision avoidance as a supervisory control problem for the DES abstraction. For solving the resulting supervisory control problem, we develop customized algorithms that exploit the structure of the transition map to compute the supremal controllable sublanguage more efficiently than standard algorithms in DES theory. We present implementation results on an intersection with several vehicles.
This is joint work with Eric Dallal at Michigan and Alessandro Colombo and Domitilla Del Vecchio at MIT.
Stéphane Lafortune has been a professor in the EECS department since 1986.