Bite Me

User interfaces have been trending towards providing more intuitive experiences with our computing devices in recent decades. Tapping, swiping, and pinching have made it simpler and faster to interact with our many gadgets. This is fantastic for most people, however, it can leave those with dexterity problems or neurological issues feeling like they are living in a world that was not designed for them. It is true that a plethora of interfaces have been designed with accessibility in mind — from voice control to eye tracking systems, and even brain-computer interfaces. But there are significant limitations with each of these systems that severely hinder their practicality under real-world conditions.

Voice controlled interfaces require a mostly noise-free environment in which to operate, and can present privacy-related concerns for their users. And of course, they also require that their user be capable of speaking clearly. Eye tracking systems require cameras and other specialized equipment, and tend to lead to user fatigue with regular use. Brain-computer interfaces are expensive, cumbersome, and more than a little bit invasive! It is quite obvious that there are many needs that are just not being fully met by the solutions available today. New options are needed, and as it turns out, a new interface that might be a better option for some has recently been described by a team of researchers from the National University of Singapore. They have developed an interactive mouthguard that controls devices with high precision through the interpretation of biting patterns with machine learning.

The smart mouthguard is equipped with an array of contact pads, each containing a phosphor that emits a different color of light when pressure is applied. When the mouthguard is bitten, the light from all of the contact pads, each producing a level of intensity that scales with the applied pressure, can be measured. To make sense of this light information, and to map it to specific patterns of biting, a machine learning algorithm was employed. This system can translate complex bite patterns into instructions for electronic devices like smartphones and wheelchairs with a 98% average level of accuracy. The technique proved to be so sensitive that it could distinguish between different types of mechanical deformations of the mouthguard, like bending, strain, and compression.

The seven gram smart mouthguard is minimally obtrusive and requires much less training to operate than most other assistive interfaces. Considering that it can be produced in single quantities in the lab for about $100, which would be expected to be much less if mass produced, the device is also very affordable. Some accommodations may need to be made to enable the mouthguard to work for a wide range of individuals in the future, however. The prototype was designed specifically for those with well-aligned teeth.

Beyond assistive technology applications, the team also see uses for their novel pressure sensors in flexible electronics, artificial skin, and dental diagnostic devices. The researchers are presently working to patent their technology and bring it to market after it has been validated clinically. They are also exploring ways to improve the processing speed of the device, and to simplify the training procedure.