Safety as a
System Property

About the Workshop

Autonomous systems have advanced rapidly and are increasingly deployed in safety-critical domains — transportation, robotics, energy, and aerospace. As these systems operate with reduced human oversight in dynamic environments and rely on learning-enabled components, failures can have severe consequences, making safety a primary requirement rather than a secondary concern.

Classical control theory provides powerful tools for stability and robustness, while advances in learning, optimization, and formal methods address uncertainty and complexity. In modern autonomous systems, however, tightly coupled perception, learning, planning, and control pipelines mean that safety cannot be fully understood by analyzing individual components in isolation.

A key insight of this workshop is that safety must be treated as a system-level property spanning the full autonomy pipeline — from data and learning to decision making, planning, and control. By bringing together ideas from robust and stochastic control, safety-critical control, distributional methods, system-level synthesis, and formal verification, the workshop aims to advance principled approaches for system-level safety assurance in data-driven autonomous systems.