ZeroBionic is a robotics company that wants to make science classrooms accessible to deaf and hard-of-hearing students using robotic arms made out of recycled plastic and locally trained sign-language data. Jamila Versi sat down with founder Maxwell Opondo to learn about the company’s origins, its ambitions, and the challenges it’s facing.
During our video call, 23-year-old Maxwell Opondo reaches away from the camera and returns holding a piece of Zerobionic’s history: the first robotic arm he deployed in a school using recycled plastic.
It is visibly rougher than the company’s current model, he says, but it carries the idea that has driven the Kenyan startup from the beginning: that accessibility and affordability need not be opposites.
Zerobionic began as a response to something Opondo repeatedly witnessed while mentoring young people in robotics, hardware and electronics. Before launching the company around two-and-a-half years ago, he worked with Young Scientists Kenya, travelling to schools and helping students develop technical projects.
In the classrooms, he noticed that deaf and hard-of-hearing pupils were often separated from activities.
Teachers told him that the scientific language was too technical to translate into sign, and in most cases, assumed the students would not be able to pursue STEM subjects at all. Opondo saw something different. “The kids were generally really interested,” he recalls.
“We have really brilliant students doing a lot of things,” he says, “but because they’re deaf, they can’t participate in a lot of these things.”
For Opondo, the need goes beyond schools. He argues that the scarcity of deaf doctors, engineers and other technical professionals is not a reflection of ability, but of systems that exclude deaf learners early.
“How many deaf or hard-of-hearing people do you know that are working in a STEM field as a doctor or an engineer?” he asks me. He predicts my answer: none.
Zerobionic concentrates much of its educational content on science subjects such as chemistry, biology, mathematics and robotics – areas where specialist sign-language vocabulary and trained interpreters can be particularly scarce.
From exclusion to invention
At first, he tried to keep teaching remotely after school visits ended. But the experiment ran into two barriers. The first was infrastructure: some schools shared limited internet bandwidth across many computers, making video teaching difficult.
The second was communication. Text-based transcription, especially for technical subjects, did not give students the language they needed.
At the same time, Opondo was researching low-cost robotic prosthetics, building on work he had done with amputees in Sierra Leone.
The two strands came together. He built a robotic arm that could communicate physically, paired it with a low-bandwidth connection and reserve power supply, and tested it with students. Pupils could see the hand move and understand the information in a way that text captions had failed to deliver.
“That is how the whole idea of using robotic arms to communicate came up,” Opondo says.
Teaching a robot how to sign
The first devices were limited mainly to finger spelling, but Zerobionic kept developing. It deployed arms in Kenya and Rwanda, while teachers and academics elsewhere began joining lessons remotely. According to Opondo, professors in London and southern Argentina were among those who used the system to share knowledge with students.
Yet finger spelling also exposed the limits of a mechanical hand. Sign language is not simply a sequence of letters spelled by hand; it uses motion across the body, and long technical explanations could become slow and cumbersome.
Zerobionic’s answer was F1, which Opondo describes as Africa’s first half-humanoid robot designed and trained specifically for sign language.
The ambition is not merely to translate speech into generic gestures. Opondo says Zerobionic is building “hyper-localised” sign-language data so the machine can adapt to regional variants. A student in Rwanda, for example, should receive communication appropriate to the sign language used there, while a teacher elsewhere can speak normally into the system.
“If you’re in Mombasa, people at the coast have a dialect. And it’s the same with sign language, so we needed a robot that could adapt to the specific dialects for different regions. Once you speak, it picks your dialect, and places you from a particular place in the world.”
That localisation is central to Zerobionic’s pitch – and one of its hardest technical problems. Many African sign languages and regional variants are poorly documented, Opondo says, leaving the company to build motion datasets while improving the robot’s dexterity. Fine joint movements remain difficult, particularly for signs that depend on subtle hand and body positioning.
Scaling beyond Kenya
The company is nevertheless scaling quickly. He was later joined by co-founder Kimathi Nora, who, also at 22, was recently awarded the Young Engineer Woman Award. The company has now grown to a team of 38 people across Africa.
Opondo says Zerobionic has made deployments in seven African countries, including Rwanda, Tanzania, Namibia, Liberia, Cameroon and Lesotho, as well as projects outside the continent in Argentina, Switzerland and Austria.
Its Nairobi operation has 12 people, while its wider African team numbers 38. He says the company reached 565,000 students across its seven African markets in the previous quarter.
Keeping accessibility affordable
The company seems almost too good to be true, I tell Maxwell. To ensure that the robotic arms remain accessible, they are offered at a scale of £50-300. The cheapest is intended as a “sponsor-funded placement for rural schools, clinics, and refugee programs”.
Affordability has shaped the engineering. When the team struggled with the cost of producing multiple arms for schools, they realised they could turn discarded plastic bottles into 3D-printing filament.
Plastic is stripped into narrow strands, heated and formed into filament, which is then fed into 3D printers to make exterior components. Opondo says the approach cuts material costs by almost 60% – although electronics remain a significant expense.
The business model is still evolving. Zerobionic generally leases its hardware and charges schools for the software platform that controls the robots, using a subscription model tailored to each environment.
The company has also explored pilots in public spaces, where sign-language translation must be trained for specific vocabularies such as airport or media terminology.
Opondo is open about the financial tension. Zerobionic is not yet comfortably profitable, he says, and is still trying to find margins that allow it to grow without pricing out the communities it was created to serve.
That tension informs its attitude to investment. The startup has taken grant and technical support, including hardware support from Nvidia and funding linked to UNDP initiatives, but Opondo says the team has deliberately tried not to depend heavily on venture capital.
“At some point for scale, we’ll need it,” he concedes, admitting that it was something they had “tried to evade”. Potential investors, he says, have to align with the company’s mission rather than push it away from low-cost accessibility or demand speed at the expense of product quality.
Growing Pains
Have they been facing scrutiny? I ask.
“I think being young in several ecosystems and also doing something that essentially is new and novel to this part of the world has to be faced by a lot of backlash,” he responds.
He recalls a demonstration in Rwanda about a year and a half ago when the robot struggled with noisy audio input. Its sign output became jumbled and produced inappropriate communication for deaf users (“Swearwords?” I asked. “Yup”, Opondo replied, both of us grinning). The episode triggered criticism and forced the team to add stricter controls around what the system can receive and output, particularly as they are working with children.
It also underlined a broader challenge familiar across artificial intelligence: models are only as good as their data and safeguards.
Now Zerobionic is concentrating heavily on that data layer. Opondo says its Nairobi lab is collecting and training motion data for multiple sign languages, while the company strengthens the server infrastructure behind the platform.
Rather than chase unlimited user growth immediately, he says it plans to hold capacity at around 3.2m students while improving the foundations needed for further expansion.
But he also sees the people Zerobionic serves not as beneficiaries, but as collaborators. After a recent deployment in Kisumu with children aged six and under, he was struck by how directly they critiqued the machines and suggested changes.
“It’s amazing how raw and unregulated kids can tell you things.” Some of the company’s improvements, he says, have come from precisely that kind of feedback.
For all the robotics and artificial intelligence behind Zerobionic, Opondo’s final argument is simple: the real problem is not deafness, but the environments people build around it. Technology can help bridge those gaps, but only if inclusion is treated as a core requirement, not an afterthought.
“We can make our spaces accessible,” he says, “without thinking they can break the bank.”
The post Kenyan inventor uses robots to open up STEM for deaf children appeared first on New African Magazine .
Kenyan inventor uses robots to open up STEM for deaf children
Aggregated summary from an independent source. Read the original at NewAfricanMagazine.