Evaluation of an ambient noise insensitive hum-based powered wheelchair controller

Abstract

Purpose: A recently-developed assistive technology nicknamed “the Hummer” was investigated as a potential powered wheelchair controller for individuals with severe and multiple disabilities. System performance in a noisy environment was compared to that obtained with a commercial automatic speech recognition (ASR) system. Method: A bi-hum driving protocol was developed to allow the Hummer to serve as a powered wheelchair controller. Participants performed several virtual wheelchair driving tasks of increasing difficulty using the two systems. Custom-written software recorded task execution time, number of commands issued and wall collisions, speed, and trajectory. Results: The bi-hum protocol was shown to be non-intuitive and required user training. Overall, the Hummer achieved lower performance relative to ASR. Once users became accustomed to the protocol, the difference in performance between the two systems became insignificant, particularly for the higher-difficulty task. Conclusions: The Hummer provides a promising new alternative for powered wheelchair control in everyday environments for individuals with severe and multiple disabilities who are able to hum, particularly for those with severe dysarthria which precludes ASR usage. A more intuitive driving protocol is still needed to reduce user frustration and mitigate user-generated errors; recommendations on how this can be achieved are given herein.

Implications for Rehabilitation

• The “Hummer” allows for noise-insensitive powered wheelchair control in everyday environments

• The “Hummer” accommodates individuals without functional speech or those with severe dysarthria to gain mobility independence, thus improving their quality of life.

Keywords::

• ambient noise
• assistive technology
• automatic speech recognition
• powered wheelchair control
• vocal cord vibration

Acknowledgements

The authors wish to thank Ka Lun Tam, Steven Xu, and Pierre Duez for hardware and software support.

Declaration of Interest: This research was funded by the Natural Sciences and Engineering Research Council of Canada and their Canada Research Chair Program. The authors report no conflict of interest.