A New Approach of Minimizing Midas Touch Problem for a Tracer-Free Tongue-Controlled Assistive Technology

Tongue-controlled assistive technologies (ATs) have a great potential to improve the quality of life for people with severe disabilities. Current tongue-controlled ATs, however, require either a tracer to be attached on the tongue or a complicated gesture/mechanism to avoid issuing commands unintendedly during routine intraoral events (e.g., tongue resting, swallowing, and speaking), which poses a high entry barrier to potential users. In this study, we propose a tracer-free tongue-controlled AT using four optical distance sensors and one ambient light sensor with a multi-sensor data fusion algorithm to distinguish the intended tongue contacts from the unintended ones. Eleven healthy human subjects participated in the study to evaluate the performance using the proposed technology in masking unintendedly-issued commands and navigating a mouse cursor on a computer. The proposed multi-sensor data fusion algorithm could mask 100%, 86.99%, and 99.50% of unintendedly-issued commands during tongue resting, swallowing, and speaking, respectively. The efficacy of the tongue commands in mouse navigation was evaluated by command-issuing task and maze-navigation task. The participants could issue tongue commands in a random order with 97.16±0.94% of accuracy. The completion time and the sum of deviation of the maze-navigation task were 31.36±2.74 s and 9.38±0.85 pixel2, respectively. The subjects' performance on computer-access tasks, using the proposed tracer-free tongue-controlled technology, is at the level of the performance measured with existing tracer-based tongue-controlled technologies, while securing robustness in masking commands, unintendedly issued during routine intraoral events. Results of this preliminary study suggest that the tracer-free implementation is feasible option for the tongue-controlled ATs.