Abstract
Background
Although many alternative visual technologies for the visually impaired have been studied, no system has been developed that provides distance information on multiple objects in front of a person with blindness while in a walking environment via electrotactile stimulation. Therefore, we studied the feasibility of a 12-channel electrotactile stimulator (Forehead ElectroTactile Stimulator [FETS]) designed for those with visual impairment. It used a stereo camera to capture depth information and adjusted the voltage output for each channel. It also used a deep-learning algorithm to identify objects and converted this information into speech.
Methods
Medical-grade SUS316 surface electrodes were mounted on a flexible silicone pad to increase the contact area with the forehead. The stimulator delivered pulsed voltages up to ± 25 V, adjustable in 2.5-V increments. The pulsed voltage parameters were set to activate A-beta nerve fibers without activating pain-related A-delta and C nerve fibers. The stereo camera captured depth information from 1 to 3 m, focusing on the proximity to the user. YOLOv8 was used to detect objects from the input video in real-time and convert the results into speech via text-to-speech (TTS).
Results
The developed FETS could output voltages from 0 to 25 V in 100 steps via Bluetooth communication, and the maximum standard deviation of the actual output voltage from the set voltage was 0.71 V. The FETS output voltage intensity was adjustable in real time according to the distance to the object captured by the stereo camera. The FETS could recognize objects in front of the user upon request and provided voice guidance through TTS.
Conclusion
We conducted this feasibility study to enhance the vision of people with visual impairment by applying a stereo camera and electrotactile stimulation to their foreheads. Information about the distance to the object acquired through the stereo camera was implemented by adjusting the output voltage according to the distance from the electrotactile stimulator. The object was identified through an artificial-intelligence algorithm and delivered by voice. However, the threshold voltage and just noticeable difference values are not the same for each person; therefore, further research is needed.