We present a method for synthesising control strategies for continuous dynamical systems. We use Uppaal Tiga for the synthesis in combination with a set-based Euler method for guaranteeing that the synthesis is safe. We present both a general method and a method which provides tighter bounds for monotone systems. As a case-study, we synthesize a guaranteed safe strategy for a simplified adaptive cruise control application. We show that the guaranteed strategy is only slightly more conservative than the strategy generated in the original adaptive cruise control paper which uses a discrete non guaranteed strategy. Also, we show how reinforcement learning may be used to obtain optimal sub-strategies.
|Konference||International Workshop on Design, Modeling, and Evaluation of Cyber Physical Systems (CyPhy 2018)|
|Periode||04/10/2018 → 05/10/2018|
|Navn||Lecture Notes in Computer Science|