Symposium on Dependability and Safety of Autonomous Systems

dummy

November 9 (Fri) 2019

19F Conference Room, National Institute of Informatics

Overview

Autonomous smart systems are emerging in various application domains, and there is a growing concern of their dependability and security. In this symposium, we invite world-leading researchers and practitioners to present their advanced experience for dependability and safety in railway, automotive, and other domains. We also have a panel discussion on challenges and future directions.

Supported by

GRACE Center ERATO-MMSD Project

Keynote Speakers

Krzysztof Czarnecki, Waterloo University, Canada

Title: Safety Assurance of Machine Learning in Automated Driving Systems

Naoki Ishihama, Japan Aerospace Exploration Agency, Japan

Title: Safety for Intelligent Space Systems

Program

This event ended successfully! Thank you for all the speakers and participants. Slides of the talks are being uploaded.

11:00-12:15 Shot-gun Talks from Researchers

13:30-14:25 Keynote Talk (1): Safety for Intelligent Space Systems (Naoki Ishihama, Japan Aerospace Exploration Agency, Japan)

14:25-15:20 Keynote Talk (2): Safety Assurance of Machine Learning in Automated Driving Systems (Krzysztof Czarnecki, Waterloo University, Canada) SLIDES (65MB)

15:20-15:45 Break

15:45-16:40 Keynote Talk (3): Integrated Approach to Dependable Cyber-Physical Systems: from Category Theory to Machine Learning (Fuyuki Ishikawa, National Institute of Informatics, Japan) SLIDES

16:40-16:50 Short Break

16:50-17:50 Panel Discussion

Invited Talks

Title: Safety Assurance of Machine Learning in Automated Driving Systems (Krzysztof Czarnecki, Waterloo University, Canada)

Automated driving systems rely on machine learning (ML) for many of their perception task, and some advocate using ML also in decision making. While ML has enabled impressive advances in performance of these tasks, assuring the safety of systems that rely on ML is a major challenge. This talk will review the key challenges that ML poses to safety assurance, describe existing strategies to address some of these challenges in practice, and briefly survey promising research directions to attack the remaining challenges.

Title: Safety for Intelligent Space Systems (Naoki Ishihama, Japan Aerospace Exploration Agency, Japan)

To appear

Integrated Approach to Dependable Cyber-Physical Systems: from Category Theory to Machine Learning (Fuyuki Ishikawa, National Institute of Informatics, Japan)

Ensuring dependability for emerging cyber-physical systems (CPS), such as autonomous driving systems, is very challenging. We obviously need advance in techniques for dependability to the increasingly complex problems, including continous dynamics, environment variability, and so on. The ERATO-MMSD Project, led by Ichiro Hasuo, put its base on "Metamathematics" for promoting theoretical extensions of formal methods (classically for software) to CPS while also involving heuristic and empirical approaches, such as search-based testing. In this talk, we show early results from the project that demonstrate the values of the heterogenous but integrated approach by combining insights from different technical areas.

Registration

Registration is closed.

Organizers

Fuyuki Ishikawa, National Institute of Informatics, Japan

Alexander Romanovsky, Newcastle University, UK

Thierry Lecomte, ClearSy, France

Contact

Fuyuki Ishikawa, National Institute of Informatics, f-ishikawa <at> nii.ac.jp