Abstract

In this talk we will demonstrate recent advances and applications of the tool UPPAAL Stratego  (www.uppaal.org) towards automatic synthesis of guaranteed safe and near-optimal control strategies for Cyber Physical Systems (CPS). The tool considers a CPS as a game between a complex (e.g. with discrete, stochastic and continuous aspects) environment and the controller.  The automatic synthesis of the tool combines symbolic methods from model checking, reinforcement learning methods from machine learning, as well as abstraction techniques for hybrid games. The talk will highlight successful applications including autonomous driving maneuvers, optimal heating systems and intelligent traffic control.