Provably Safe Reinforcement Learning via Action Projection Using Reachability Analysis and Polynomial Zonotopes

While reinforcement learning produces very promising results for many applications, its main disadvantage is the lack of safety guarantees, which prevents its use in safety-critical systems. In this work, we address this issue by a safety shield for nonlinear continuous systems that solve reach-avoi...

Ausführliche Beschreibung

Bibliografische Detailangaben

Hauptverfasser:	Niklas Kochdumper, Hanna Krasowski, Xiao Wang, Stanley Bak, Matthias Althoff
Format:	Artigo
Sprache:	Inglês
Veröffentlicht:	IEEE 2023-01-01
Schriftenreihe:	IEEE Open Journal of Control Systems
Schlagworte:	Action projection reach-avoid problems reachability analysis reinforcement learning
Online-Zugang:	https://ieeexplore.ieee.org/document/10068193/
Tags:	Tag hinzufügen Keine Tags, Fügen Sie das erste Tag hinzu!