Home // ADAPTIVE 2022, The Fourteenth International Conference on Adaptive and Self-Adaptive Systems and Applications // View article

Qualitative Monitors based on the Connected Dependability Cage Approach

Authors:
Felix Helsch
Iqra Aslam
Abhishek Buragohain
Andreas Rausch

Keywords: Dependable Autonomous System; Connected Dependability Cage; Runtime System Observation; Resilience System Stabilization; Qualitative Monitoring

Abstract:
Many Autonomous Systems (ASs) have been widely applied in safety-critical applications like driverless taxis and financial credit assessment. Due to the integration of machine learning techniques for functions like environmental perception, ASs are nowadays a hybrid construction combined with classical engineered and Artificial Intelligence (AI-) based subsystems. Such a construction of the hybrid AI-based AS makes it impossible for engineers to guarantee the dependability requirements during development, since the system cannot be completely tested and formally verified. Addressing these dependability issues of the hybrid AI-based AS, this paper provides a transparent overview of our Dependability Cage approach on different levels of abstraction. In particular onboard continuous monitoring is com bined with remote technical supervision for human intervention in our approach for Runtime System Observation and Resilience System Stabilization, forming a Connected Dependability Cage. The Qualitative Monitor for observing the system’s functional correctness at runtime is chosen as an implementation example and is evaluated in the concrete use case of the research project ”VanAssist” which focuses on using AVs for package delivery in urban areas.

Pages: 46 to 55

Copyright: Copyright (c) IARIA, 2022

Publication date: April 24, 2022

Published in: conference

ISSN: 2308-4146

ISBN: 978-1-61208-951-5

Location: Barcelona, Spain

Dates: from April 24, 2022 to April 28, 2022