Young Researchers Symposium 2026
The Predictive Robotic Mind
Why Robots Should Learn to Be Surprised
Can robots use experience not only to remember the past, but to form expectations about what happens next — and recognize when reality no longer matches them?
Abstract
From experience to expectation
Humans adapt not only by perceiving the world, but also by comparing what they perceive with what they expected. When the world behaves differently than expected, we become cautious and adjust our actions. Autonomous robots still struggle with this kind of self-assessment: they can perceive their surroundings, but often cannot recognize when their interpretation may be uncertain.
This work proposes the Predictive Robotic Mind, a human-inspired concept for adaptive autonomy. It compares current observations with memory-based expectations and uses prediction error as an internal signal for behavior adaptation. The goal is to develop robots that not only detect their surroundings, but also recognize when the world no longer matches their expectations.
About the author
Catharina Helten
My research explores how ideas from cognitive science can help robots use memory, expectations, and uncertainty to adapt in changing environments. Before starting my PhD at RPTU, I worked as a safety engineer at Fraunhofer IESE, which continues to shape my focus on reliable and understandable autonomy.
CH