Post-conference Workshop Day | Thursday, 2 April 2020

7:30 am - 8:00 am Welcome coffee and registration for workshop 6 and 7

8:00 am - 10:30 am WORKSHOP 6 - Safety assurance for Machine Learning

This workshop will explore approaches to assuring the safety of the use of machine learning. It will consider the issues across the machine learning lifecycle, looking at data management, model learning, model verification and model deployment.

8:00 am - 10:30 am WORKSHOP 7 - Validating safe functionality - handling unknown unsafe scenarios

Focus: Validation according to ISO 21448:2019 SOTIF

Verification and validation are the key activities in ensure safety of the intended function. While verification focusses on the process of rendering known unsafe scenarios safe, validation is the part that is looking into the unknown unsafe scenarios. In this workshop, we will discuss both verification and validation strategies and then concentrate on validating the SOTIF.
• Discuss with your peers different verification and validation strategies and exchange on lessons learnt.
• Get an understanding on how to identify potential triggering conditions on the left side of the V-model
• Improve your validation process by learning from Best Practice examples

10:30 am - 11:00 am Coffee break and registration for workshop 8

11:00 am - 1:30 pm WORKSHOP 8 - An introduction to STPA (System Theoretic Process Analysis)

John Thomas - Staff Researcher, Massachusetts Institute of Technology
The system theoretic process analysis (STPA) is a much discussed risk and hazard analysis method. As a top-down analysis, STPA emphasizes the system‘s dynamic behavior including automation interactions and human behavior. Join this workshop to get a hands-on approach how to perform STPA and it’s relationship to the safety of the intended functionality.
• Get a summary of the most important aspects of the STPA: basic concepts, terminology, and process will be explained
• Understand the difference between traditional techniques and STPA, and latter‘s advantages
• Apply an exemplary STPA to a limited case study to demonstrate the process

John Thomas

Staff Researcher
Massachusetts Institute of Technology

1:30 pm - 2:30 pm Networking luncheon and registration for workshop 9 and 10

2:30 pm - 5:00 pm WORKSHOP 9 - Safety Architectures and ASIL Decomposition

The workshop will describe the importance of defining safety architectures for high-performance SoCs targeting high-ASIL by leveraging from ASIL decomposition. Topics that will be included are:
• ASIL decomposition as powerful technique
• HW/SW architecture options
• Dealing with legacy HW/SW and QM components

2:30 pm - 5:00 pm WORKSHOP 10 - Keeping AI Safe - Specification and Robustness for Building Safe Artificial Intelligence

Stefan Milz - Research Fellow - Department of Computer Science and Automation, Ilmenau University of Technology
Focus: Safe AI

The higher we climb up the ladder of automation towards levels 3, 4 and 5, the less we can leave the issue of safe AI aside.
In this workshop, we want to explore how we can build Artificial Intelligence systems in such a way that they are robust and safe.
 • Understand the challenges in AI for safety critical systems
• Get an overview of ongoing and future initiatives on making AI safe
• Share your experience on how to specify and build robust AI
• Discussion on the possibility of including a guideline in the SOTIF on how to increase the safety of the AI

Stefan Milz

Research Fellow - Department of Computer Science and Automation
Ilmenau University of Technology

5:00 pm - 5:00 pm End of the 2020 Safety Week