smartARM Builds AI Prosthetic That Recognizes Objects on Its Own

A Toronto startup is developing a bionic arm that uses artificial intelligence to make prosthetic control more intuitive for people with limb differences.

smartARM's vision-first prototype relies on open-source software, with Meta AI glasses offered as an optional add-on. The arm "sees" through a camera embedded in its palm, using Meta's open-source vision model DINOv2 to recognize everyday objects from just a few reference photos.

Traditional prosthetic hands often require users to switch manually between grip patterns, turning simple tasks like picking up a glass and then a spoon into a multistep process. smartARM's grip-selection software reads visual features directly and chooses a suitable grip automatically, without prior understanding or extra programming. Objects that once took weeks to train can now be learned almost instantly.

The Meta AI glasses and Meta Wearables Device Access Toolkit add egocentric data, giving the arm more context about what the user wants to interact with. Users can also add new objects through a phone app, letting the smartARM community personalize their prosthetics.

Long-time user and former NFL player Shaquem Griffin said scalability and accessibility matter most to him: "This is the future of consumer prosthetics." Founder and CEO Hamayal Choudhry said building a capable hand is only part of the challenge. "Making it intuitive to use matters just as much," he said, adding that the goal is an experience where people focus more on what they want to do and less on how to operate their prosthetic.