Maia Stiber (John Hopkins)- Towards Human-Aware Robots: Social Signals for Robot Error Detection in HRI

Abstract: Effective human-robot interaction relies on a robot’s ability to interpret and adapt to human behavior in real-time. My research advances human-aware robots by using social signals as inputs for understanding human internal states and models. Focusing on error-awareness, I develop a flexible, social signal-based framework that detects robot errors through natural human reactions, enabling error detection across tasks and users. To build on this, I introduce a proactive, multimodal system that actively involves users in error detection, promoting reliable and timely error detection. Through my research, I take a step towards deploying human-aware robots that can adapt to user behaviors, improving both the reliability and acceptance of robotic systems in everyday environments.