MR-Sense: A Mixed Reality Environment Search Assistant for Blind and Visually Impaired People

Abstract

Search tasks can be challenging for blind or visually impaired people. To determine an object's location and to navigate there, they often rely on the limited sensory capabilities of a white cane, search haptically, or ask for help. We introduce MR-Sense, a mixed reality assistant to support search and navigation tasks. The system is designed in a participatory fashion and utilizes sensory data of a standalone mixed reality head-mounted display to perform deep learning-driven object recognition and environment mapping. The user is supported in object search tasks via spatially mapped audio and vibrotactile feedback. We conducted a preliminary user study including ten blind or visually impaired participants and a final user evaluation with thirteen blind or visually impaired participants. The final study reveals that MR-Sense alone cannot replace the cane but provides a valuable addition in terms of usability and task load. We further propose a standardized evaluation setup for replicable studies and highlight relevant potentials and challenges fostering future work towards employing technology in accessibility.