Masterclass: Embodied AI Safety Principles for Automated Driving

Location: Workshop Room III (onsite), Hofburg, Vienna
Date: September 17, 2025
Time: 16:00 – 17:30 CEST

Host

Philip Koopman, Faculty Emeritus, Carnegie Mellon University

Masterclass Description

Automated Driving (AD) system safety goes far beyond fault tolerance and software validation. This session introduces key insights from Prof. Koopman’s upcoming book Embodied AI Safety (September 2025), with direct relevance to AD technologies.

Achieving safety at scale demands expertise in four domains: System Safety, Cybersecurity, Machine Learning, and Human–Computer Interaction. These skill areas form the foundation of trustworthy autonomous driving systems.

The workshop will explore:

• Why traditional safety frameworks fall short in AD applications
• How Embodied AI reshapes safety engineering principles
• Real-world challenges and the limitations of current standards
• New approaches to framing acceptable risk
• Pathways to building public trust through continuous safety improvement

Masterclass Overview

1. Welcome & Introduction (5 min) – Kickoff and overview of session goals
2. Keynote Presentation (25 min) – Exploring core themes including safety, security, human-machine interaction, machine learning, liability considerations, and new approaches to safety engineering and building trust
3. Live Q&A Session (10 min) – Direct interaction with the speaker on the masterclass topics
4. Interactive Workshop (45 min) – Three guided discussions applying principles from the talk to real-world challenges
5. General Q&A (up to 15 min) – Additional time for open questions
6. Closing Remarks (5 min) – Final thoughts