In a classroom, a teacher speaks and a deaf student waits for someone to translate. In much of Africa, that someone does not exist. Norah Kimathi, a 22-year-old Kenyan computer scientist, is building a fix.
Kimathi co-founded ZeroBionic, a Nairobi start-up that has built a robotic hand able to translate a teacher’s spoken words into sign language in real time. A deaf child can then follow the lesson alongside hearing classmates, without a dedicated human interpreter in the room.
A gap in the classroom
The need is large. According to the Kenya Society for Deaf Children, about 300,000 children in Kenya have a hearing impairment, yet only around 20,000 were enrolled in school as of 2024. Teachers who sign are scarce, and so are interpreters.
ZeroBionic’s answer is practical. Each hand is 3D-printed from recycled plastic and sells for about $350. It is reported to run for roughly three years without maintenance, and the company has already sold 78 units. Kimathi says the system reaches 92 percent speech-to-sign accuracy.
Starting from scratch
The hardest problem was not the hardware. “In Africa we don’t have a standardised African sign language dataset,” Kimathi told AFP. Nobody on the continent had built a system like this, so her team had to start from nothing.
ZeroBionic is building the dataset itself. At Kasarani Treeside Secondary School for the Deaf in Nairobi, deaf teachers wear a motion-capture bodysuit that records their signs, and the team stores them in a growing database of technical terms. These draw on English, American and Kenyan signs, and the database could make it easier for deaf students to learn and talk about subjects like science.
For students like 19-year-old Sharon Mumbe, the technology offers something beyond convenience. She says it gives her hope for the future.
What comes next
The team is also working on Africa One, a humanoid robot built in-house, to refine and expand the signs its mechanical arms can produce. Sign language relies on more than hands, including facial expression and body movement, so a fuller robot should widen the vocabulary it can express.
Why it matters beyond Kenya
Nigeria faces a similar problem. Deaf learners here also depend on too few interpreters and specialist teachers, and African sign languages remain poorly documented in digital form. ZeroBionic shows that a young African engineer can build technology around African realities, using local data, local materials and a price schools can afford.
It also shows the limits of imported solutions. Off-the-shelf AI tools were not built for Kenyan Sign Language, so Kimathi had to build the data first.
The question for the rest of the continent is who will do the same for Nigerian Sign Language, and how soon.
