Keynote speakers

Simon Burton

Robert Bosch GmbH, Stuttgart Area, Germany

Title

Challenges and advancements in arguing the safety of autonomous driving

Abstract

Over the last few years, there has been increasing hype surrounding the topic of automated driving and associated technologies such as machine learning. These technologies have the potential to radically decrease the number of road accidents as well as introducing significant convenience and ecological benefits. Although much technical progress has been made, the challenges of validating the safety of such systems is delaying their widespread introduction. This presentation describes the difficulties in arguing the safety of automated driving systems and the use of machine learning in particular. A framework is presented for creating an assurance case for such systems and a summary of recent advancements in this area is given. The presentation covers safety assurance from a systems engineering perspective and also addresses how such a perspective can be used to support wider ethical discussions associated with the technologies. The talk will conclude with an outlook on further activities both within academic research as well as industrial collaboration that are required in order to create a sufficient level of trust in these paradigm changing technologies.

Short Biography

Dr. Simon Burton graduated in computer science at the University of York, where he also achieved his Phd on the topic of the verification and validation of safety-critical systems. Dr. Burton has a background in a number of safety-critical industries. He has spent the last 18 years mainly focusing on automotive, working in research and development projects  as well as leading consulting, engineering service and product organisations providing solutions for process improvement, embedded software development, safety and security. He currently has the role of Chief Expert at Robert Bosch GmbH, where he coordinates technical strategy in the area of vehicle computers, software, safety and security. In addition, he has the role of visiting research fellow on the Assuring Autonomy International Programme at the University of York.

 

 

 

Alexander Romanovsky

Newcastle University, Newcastle upon Tyne, UK

Title

Holistic Fault Tolerace

Abstract

The existing approaches to ensuring fault tolerance typically rely on recursive system structuring or system layering, with the units of system structuring viewed as error confinement and error recovery areas. The aim is to reduce system complexity by realising the divide and conquer principle. We discuss various limitations of these approaches, including the fact that the developers of individual components (units) have to make worst-case assumptions about the design of the rest of the system, and the difficulties in expressing system-level fault tolerance strategies that directly involve any component or several components in recovery after an error is detected. When such approaches are applied, the developers do not have suitable means available to them to reason about various system-level concepts in the context of fault tolerance (e.g. power consumption and performance). Moreover, these approaches may complicate fault tolerance maintenance, reuse, modification and adaptation as fault tolerance can be scattered over a number of components. By contrast, holistic fault tolerance complements the existing structuring approaches by supporting a cross-cutting development of fault tolerance that addresses these issues. It allows system designers to express a system-wide view on the best way of dealing with a given fault and/or error.  Approaches like this could, however, potentially increase system complexity if applied in an undisciplined fashion. To prevent this, holistic fault tolerance relies on introducing dedicated structuring units that define fault tolerance behaviour. In addition, a number of approaches to holistic fault tolerance engineering (including architectural design, modelling, and implementation) have been proposed and evaluated in the development of various many-core and mobile applications. Finally, tolerating cross-boundary and emerging failures in systems of systems is identified as another area where the ideas of holistic fault tolerance could be successfully applied.

Short Biography

Alexander Romanovsky is a Professor of Computing Science with the School of Computing, Newcastle University, UK. He received a MSc degree in Applied Mathematics from Moscow State University and a PhD degree in Computer Science from Saint Petersburg State Technical University (Russia). He has been with Newcastle University since 1996. His main research interests are system dependability, fault tolerance, safety, software architectures, exception handling, error recovery, system verification for safety, system structuring and verification of fault tolerance and safety. Since 1992 Prof Romanovsky has been involved in a number of EC and UK research projects on various aspects of system dependability engineering. He has coordinated two major ICT projects, the RODIN STREP and DEPLOY IP, and has been the Principle Investigator of the TRAMS-2 and STARTA EPSRC/UK platform grants. He is now working on developing scalable solutions for automatic verification of signalling safety in close cooperation with the railway industry. Prof Romanovsky is a visiting professor at the National Institute of Informatics (NII, Tokyo, Japan).

 

 

 

 

Lydia Chen

Lydia Chen

TU Delft, Delft, Netherlands

Title

Machine Learning for Dependability Management

Abstract

While machine learning (ML) models are becoming the core component in today's dependability management. new dependability changes emerge on such data-driven approaches, e.g., dirty data attacks. In this talk, I will first present how the dependability management, particularly cloud data center service outages, evolve with different ML models. Secondly, I will show the vulnerabilities of those model assumptions and how to leverage the principle of adversarial learning to design robust ML-based dependability management systems. 

Short Biography

Lydia Y. Chen is an Associate Professor in the Department of Computer Science at the Delft University of Technology in The Netherlands. Prior to joining TU Delft, she was a research staff member at the IBM Research Zurich Lab from 2007 to 2018. She holds a PhD from Pennsylvania State University. She was a co-recipient of the best paper awards at CCGrid’15 and eEnergy’15. She received TU Delft technology fellowship in 2018. She has served on the editorial boards of IEEE Transactions on Distributed and Parallel Systems, IEEE Transactions on Service Computing and IEEE Transactions on Network and Service Management. She is a Senior Member of IEEE.