Psyonic wants bionic hands to help teach robots how to grip

Psyonic is exploring a plan to put thousands of bionic hands into everyday use, then use the resulting movement data to train robotic hands. The San Diego startup’s idea could turn ordinary tasks such as gripping and lifting into a new source of data for a long-standing robotics problem.

Watch Desk analysis

What happened

Psyonic chief executive Aadeel Akhtar told Semafor the company is in the early stages of an effort to deploy the hands globally, with Pakistan and India among its targets. Users would go about their daily tasks while data from their grips, lifts and other hand movements is used to train a robotics model. The hands can also sync with a wearable camera, so the system can learn what a task looks like as well as how much force it takes. Semafor’s report describes the proposal; it is an early-stage effort, not a completed global rollout.

Why it matters

Robotic hands remain difficult to build because they must handle objects and tasks that vary in shape, weight and feel. Psyonic’s approach would gather data from people using bionic hands in everyday life, rather than relying only on controlled demonstrations. That could give robotics developers examples of both the visual task and the force involved.

It also makes the people wearing the devices part of the data supply chain. Akhtar says Psyonic intends to pay participants, a detail that matters when personal activity is being turned into training material.

Our read

This is a promising answer to a stubborn robotics-data problem, and a more grounded one than hoping a robot hand will learn the world from a tidy laboratory exercise. But the idea’s success will depend on what data is collected, how users control its use and whether everyday examples actually help robots handle unfamiliar tasks. For now, treat it as a plan with potential, not proof that dexterous robot hands are solved.

What to watch

• Whether Psyonic moves from early planning to a defined deployment, and how many users take part.
• What data the hands and wearable cameras collect, and how consent, access and payment work.
• Whether the resulting data improves robotic manipulation outside the tasks used to collect it.

Discussion spark: If bionic-hand users help generate data for robotic hands, what should they be able to control beyond receiving payment?

Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.