Abstract

The white cane has been the prominent and widely used mobility aid by visually impaired persons for many years, however, there are some limitations associated with the white cane mobility aid device. Primarily, the white cane exhibits restricted capability in detecting ground-level obstacles in proximity and does not provide reliable detection of aerial obstacles. Our work proposes a device for the safe navigation of visually impaired persons utilizing a non-invasive galvanic vestibular stimulation (GVS) technique. By delivering a signal (1-1.2mA) delivered behind the ear via electrode pads to the vestibular system, we induce a sensation of steering thereby facilitating navigation. Our proposed device utilizes a combination of Intel’s D435 depth camera and an object detection model, YOLOv5, to identify and process the detection of obstacles within a 3-meter range. Within 0.2 seconds, the object is detected, and the algorithm assesses the situation and sends instructions via User Datagram Protocol (UDP) packets wirelessly to the GVS device. The device receives the packets and steers the subject (human) autonomously. In total, four tests with different scenarios have been conducted, through experimentation, it was found that the system could successfully detect, process, and transmit geospatial information to the GVS module; and steers the user into the correct trajectory to avoid any hazards obstructing the user’s path.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call