AI-Powered Electrical Stimulation Guides New Human Movements

Researchers at the University of Chicago have developed a new system that combines artificial intelligence with electrical muscle stimulation (EMS) to physically guide human movements through unfamiliar tasks. This system, described by the authors as embodied AI, aims to provide general-purpose, context-aware assistance by directing a user’s muscles to perform specific motions.

The project was led by PhD students Yun Ho and Romain Nith, under the supervision of associate professor Pedro Lopes in the Department of Computer Science. Their work received the Best Paper Award at the ACM CHI 2026 conference.

Moving Beyond Specialized EMS

Electrical muscle stimulation has been utilized for several years in various capacities. Previous applications included delivering controlled currents to help users learn piano sequences, demonstrate sign language, or assist in stroke rehabilitation.

However, the research team noted that previous EMS systems functioned primarily as training wheels. These earlier iterations were highly specialized and fixed, meaning they were only effective for a narrow set of programmed tasks and could not adapt to unpredictable real-world environments.

The University of Chicago team describes previous assistance as non-contextual. In these older systems, muscle instructions only matched the specific situations anticipated by the designer. For example, a device programmed to help a user shake a can of spray paint would be unable to recognize when a user is holding a can of cooking oil, which does not require shaking.

Context-Aware Guidance

The new system shifts toward context-aware assistance, allowing the AI to understand the environment and the specific object being handled. This capability allows the system to guide users through a variety of unfamiliar physical tasks.

Potential applications for this technology include guiding a person through the motion of opening an unfamiliar window in a foreign country, twisting open a child-proof pill bottle, or operating a camera for the first time.

By integrating AI with EMS, the system can provide a sensation where the user’s own muscles are gently guided through the necessary motions, effectively acting as an invisible teacher providing physical know-how.

Broader Context of AI in Muscle Stimulation

The integration of AI and electrical stimulation is also being explored in clinical rehabilitation settings. A separate study published on September 9, 2025, discussed NeuroSkin®, an AI-driven wearable functional electrical stimulation (FES) system designed for post-stroke gait recovery.

The NeuroSkin® system is designed to be integrated into routine stroke rehabilitation with minimal training. It enables real-time, multi-muscle stimulation to assist patients in recovering their walking abilities.

While the University of Chicago’s work focuses on general-purpose guidance for unfamiliar tasks, these parallel developments in FES highlight a growing trend of using AI to make muscle stimulation more adaptive and responsive to the user’s immediate physical needs.