BabbleMind: reducing nonverbal ambiguity via inner voice visualization in interaction with a baby-type robot

Therapeutic robots are widely used in eldercare to alleviate loneliness and support psychological well-being. Among them, baby-shaped robots are well accepted due to their familiar appearance. Their interactions mainly rely on nonverbal cues such as gestures and vocalizations. However, users who struggle to interpret these cues may have difficulty understanding the robot's intentions, reducing interaction quality. To address this issue, we propose BabbleMind, which visualizes a baby robot's intentions (inner voice) on an onboard display. BabbleMind combines Semantic Babbling with explicit inner-voice visualization to support intention understanding without altering nonverbal behavior. Experimental results showed that inner-voice visualization enhanced empathy, immersion, and trust. A case study with older adults further suggested improved intention understanding and interaction ease.