Russian Scientists Build Smarter Eye-Tracking Algorithm for Assistive Tech

A new machine learning model developed in Moscow could reduce false activations and boost accuracy in eye-controlled computer systems.
Russian researchers have created an innovative machine learning algorithm that improves the accuracy and usability of computer systems controlled by eye movement.
Eye-tracking technologies are already being used in virtual and augmented reality, as well as in assistive tools for people with motor impairments. But existing systems often misinterpret user intent, triggering unwanted actions due to involuntary gaze behavior.
To address this, the new algorithm distinguishes between accidental gaze delays and intentional commands. Researchers discovered that unintentional pauses tend to last around 500 milliseconds and exhibit unique characteristics. The model incorporates two classifiers: one that tracks micro-movements of the eyes, and another that interprets screen context—such as object locations or game strategy.
Tests showed the system reduced false activations by a factor of three. Developed by scientists at Moscow State University of Psychology and Education, the project has already received a grant from the Russian Science Foundation, signaling official recognition as a technological breakthrough.