EgoThumb uses a camera to decide when an extra thumb should grip

Reaching for a bottle can be enough to set EgoThumb to work: the added robotic digit holds the container while the wearer's natural fingers turn its cap, then lets go when it is set down. The interesting step is not simply adding another finger. It is shifting some decisions about grasping and releasing from the person to a visual control system. The research became public on October 27, 2026, and remains a laboratory prototype.

Extra-thumb experiments already have a substantial history. Dani Clode's Third Thumb sits opposite the natural thumb and responds to pressure sensors beneath the big toes. UCL reported in 2021 that trained participants could perform one-handed tasks such as building block towers and carrying several objects. Some could operate it while solving a math problem. Having an additional digit, or learning to coordinate it with the hand, is not new.

A separate Third Thumb study published in 2024 tested 596 people aged three to 96. That broad experiment examined how readily different users could take up body augmentation. EgoThumb asks a different question: could the machine take over part of the control burden instead of requiring a separate command for each action?

Researcher Jiawei Fu told Tech Xplore that soldering wires inspired the project, although soldering itself has not been tested. The team began with simpler holding tasks. A person first provided demonstrations through a foot-operated controller; the trained system then used camera images and joint positions to plan movements. The distinction resembles an assistant watching a partner's actions and anticipating when to help, rather than a pedal that moves only when explicitly pressed.

The original paper describes three users, with ten trials per task for each. Expanding the area where objects were placed reduced average success on the ball-transfer task to about 17 percent. Computation also ran on an external computer. Those details put a useful boundary around the demonstrations: successful cooperation in a familiar arrangement is not dependable assistance wherever a person moves their hand.

The everyday appeal is easy to picture. When one hand is occupied fixing something in place, an additional grip could free the other from that supporting job. But a wearable helper must know when to hold and when to release, not merely reproduce an impressive isolated demonstration. EgoThumb offers a concrete approach to sharing the coordination work. Reliable assistance across people and situations remains the test that matters beyond the laboratory.

Original paper posted: October 28, 2026, 01:22 Beijing time / October 27, 2026, 1:22 PM EDT.