Hands-free locomotion is essential for VR, yet existing hardware-based solutions remain costly and inaccessible. While voice interaction offers a lightweight alternative, existing approaches rely on rigid commands and struggle with ambiguity. We first conducted a Wizard-of-Oz study with 16 participants to investigate natural multimodal communication patterns. Based on the findings, we developed Dialo-go, an LLM-powered locomotion system that integrates voice and gaze through a context-aware pipeline for robust intent interpretation. We evaluated Dialo-go in two within-subject studies (goal-oriented navigation and free exploration) against two system variants: a voice-based teleportation system and a gaze-enhanced teleportation system. Results demonstrate that Dialo-go significantly improves task performance and overall user experience. Our work presents a practical architecture for conversational multimodal VR locomotion and provides empirical insights and design implications for developing natural, robust, and accessible hands-free interaction.
Zheng et al. (Wed,) studied this question.