H-STAR2026: Human State‑Aware Robotics: From Multimodal Data to Human‑Adaptive Behavior in HRI

Recent advances in sensing technologies enable robots to capture rich multimodal human signals, including gaze, speech, motion, and physiological responses. However, most systems still rely on event-driven behaviors and predefined policies, limiting their ability to incorporate continuous estimates of human internal states into real-time control. This workshop addresses the integration of multimodal human state estimation with robotic decision making. It explores approaches that connect sensing, representation learning, and adaptive planning across cognitive and social dimensions of interaction, with a focus on dynamic states such as engagement, anxiety, and trust. By covering the full pipeline from data curation and multimodal fusion to real time deployment, the workshop also highlights challenges including noise, latency, and ethical concerns, aiming to advance robust, context aware, and human centered HRI systems.