Guide Dog Robot for Visually Impaired People: Audio Source Localization and Classification for Hazard Detection

Presenter

Shiven Patel

Campus

UMass Amherst

Sponsor

Hochul Hwang, Department of Computer Science, UMass Amherst

Schedule

Session 2, 11:30 AM - 12:15 PM [Schedule by Time][Poster Grid for Time/Location]

Location

Poster Board A64, Campus Center Auditorium, Row 4 (A61-A80) [Poster Location Map]

Abstract

The integration of robotic guide dogs provides an alternative mobility aid to the blind and low-vision individuals unable to work with traditional animal guide dogs, offering benefits in cost-effectiveness, potential for widespread distribution, and minimal maintenance. Despite advances in robotic guides, challenges exist in detecting potential hazards on sidewalks (e.g., electric scooters, bicycles) relying only on visual information. This includes difficulties in capturing visuals in extremely bright or dark conditions and occlusions for the camera that are inevitable when the human-robot closely collaborates.

Addressing this issue, my research introduces an innovative approach by employing a microphone array for audio localization and classification to enhance the sensory capabilities of guide dog robots making them truly multi-modal. This method aims to accurately estimate the azimuth and distance of sound-emitting environmental threats, such as oncoming vehicles, electric scooters, and bicycles.Our methodology leverages the temporal information obtained from sound to complement the visual data, creating a more comprehensive perception system for the robot. By integrating audio cues into the path planning algorithm, the robot can navigate more safely and efficiently, providing a superior guidance experience for its user. 

This advancement in guide dog robots not only broadens the horizons for assistive technology for the visually impaired but also contributes to the wider field of robotics by highlighting the importance of multimodal sensory integration for overcoming environmental challenges. My research underscores the potential of audio localization as a critical tool for enhancing the capabilities of robotic navigation systems in complex and dynamic landscapes.

Keywords

Audio Localization, Multi-Modal Hazard Detection, Guide Dog Robot, Audio Classification, Visually Impaired

Research Area

Engineering

SIMILAR ABSTRACTS (BY KEYWORD)