Presentation Information
[C-12-44]Implementation and Evaluation of a Visual-to-Auditory Conversion Module in a Wearable Navigation System for Visually Impaired People
◎Kotaro Torazawa1, Aoi Kataura1, Duan Yuqi1, Akihiro Suzuki1, Ryotaro Kawasima1, Pattaramon Thianmontri1, Yukino Sakurai1, Takafumi Fukusima1, Koji Kiyoyama2, Tetsu Tanaka1 (1. Tohoku Univ., 2. Nagasaki Inst of Applied Science)
Keywords:
Navigation system,visual-to-auditory conversion,object detection,walking support
For safe walking by visually impaired users, assistive systems using GPS, maps, cameras, and distance sensors have been proposed. They detect obstacles and notify users by vibration or sound. However users need to interpret the stimuli before avoiding obstacles, causing delayed responses and cognitive load. To address this issue, this study proposes a wearable navigation device, “SMC-Device,” combining a Sensory Modality Conversion (SMC) module and a Galvanic Vestibular Stimulation (GVS) module. This paper focuses on the SMC module, which converts visual information into sound. The module recognizes pedestrians, vehicles, traffic lights, and other objects in real time using YOLOv11. Object category is mapped to sound pitch, while distance is mapped to sound volume, allowing users to intuitively perceive object type and proximity. The sound generation circuit consists of a DAC, VCO, LPF, and VGA, and was fabricated using TSMC 0.18-µm CMOS technology. Evaluation confirmed stable circuit operation, distance-dependent volume changes, and frequency changes during target switching in multi-object environments. The power consumption was 7.6 mW, indicating suitability for wearable devices.