Wheelie 7 – Wheelchairs with Human Expression Control
Brazilian startup Hoobox Robotics is producing an adapter kit that allows electric wheelchairs to be controlled by the users’ facial expressions, therefore taking advantage of the user’s best abilities to counteract their physical limitations.
About 10% of the global population – almost 650 million people, have disabilities, and of which, studies indicate, some 10% require a wheelchair. Following a mobility-related injury, many people undergo significant emotional and mental trauma along with the more obvious physical impairment. Hoobox claims to not only improve this mobility through the Wheelie 7, but also enhance users’ self-esteem using this tool.
Wheelie 7, so called because it takes only seven minutes to install, uses Intel’s AI 4 Good programme – Intel’s 3D RealSense Depth Camera and a mini onboard Computer, combined with Hoobox’s facial recognition software. The Wheelie 7 kit equips a wheelchair with AI to detect 9 user facial expressions and processes the data in real-time to direct the movement of the chair. Smiling, raising the eyebrows, wrinkling the nose or pouting the lips as if for a kiss are among the repertoire of gestures recognised by the prototype Wheelie 7.
The kit detects precise human behaviour such as drowsiness, 10 levels of pain, agitation or sedation levels and spasms. The customisation of the expressions to movements is simple with almost no training required and with immediate effect.
Hoobox CEO and cofounder Paulo Gurgel Pinheiro says, “We have mixed a creative idea with the right hardware and the right AI platform, so that when we put it together, we can make this great tool for giving autonomy and mobility back to the users”.
The Wheelie 7 prototype is currently being tested by users in the US, including people with quadriplegia and motor neurone disease, as well as seniors. In March 2019, Wheelie 7 will be made available commercially with a subscription fee of $300 per month. 400 kits of the adapter will be released as a first batch.