Robin J. Kirschner presents Safety Certification for Autonomous Robots: Challenges in Human-Robot Interaction

Ensuring safe human–robot interaction is one of the defining challenges of our
transition into a shared‑autonomous world. Demonstrating that robotic systems
behave safely is far from trivial, and several critical research gaps continue
to limit our ability to build and certify trustworthy robots. First, the lack of
comprehensive human injury data constrains a robot’s understanding of what
constitutes safe operation across diverse real‑world scenarios. Second,
incomplete or imprecise robot models hinder the meaningful integration of this
safety knowledge into the robot’s own planning and decision‑making
processes. Finally, we face a deep technical challenge: developing sensing and
control systems that are not only reliable, but whose reliability can be
rigorously demonstrated. In this talk, we examine these three core obstacles to
robotic safety, explore how they shape the current landscape of human–robot
interaction, and outline emerging ideas and research directions aimed at
overcoming them. Together, these threads form the foundation for a future in
which robots can operate safely, intelligently, and collaboratively alongside
humans.