A Visualization-Based Approach to Bodily Sensation Awareness using AR and LLM-generated Meditation Guidance

초록

This work presents an augmented reality (AR) meditation system that visualizes internal bodily sensations through circular objects mapped to key body joints. Using MediaPipe, the system tracks pose and hand gestures to enable users to adjust chakra-like visual elements via pinch gestures, reflecting changes in bodily attention. A local large language model (LLM, Ollama) generates meditation prompts, which are then converted to speech via TTS. These components create a layered experience in which the body informs visual interaction and language guidance supports sustained inward focus, offering an embodied meditation environment that bridges somatic awareness with reflection. © 2025 IEEE.

키워드