11.10.2024

A research team led by EPFL Professor Alexander Mathis has developed an AI model that provides in-depth insights into hand movements, crucial for advancing neuroprosthetics and rehabilitation technologies.

In neuroscience and biomedical engineering, accurately modeling the complex movements of the human hand remains a major challenge. Many existing models fail to capture the intricate connections between the brain’s motor commands and the physical actions of muscles and tendons. This gap has implications for both scientific research and the development of neuroprosthetics for patients with limb loss or paralysis.

Image: AI model of a hand to research motor manipulation; Copyright: EPFL / UNIGE

The EPFL/UNIGE AI model achieved a 100 percent success rate in the first phase of the competition, surpassing the closest competitor. Even in the more challenging second phase, their model showed its strength in ever more difficult situations and maintained a commanding lead to win the competition.

Products and exhibitors related to this topic

Exhibitors and products related to this topic can be found in the database of MEDICA:

A new AI approach to understanding motor control

Prof. Alexander Mathis and his team at EPFL developed a novel AI-based approach, advancing the understanding of motor functions. By employing a combination of curriculum-based reinforcement learning and biomechanical simulations, the research provides a dynamic and anatomically precise model of hand movements, offering a new perspective on how humans learn complex motor skills. “What excites me most about this research is that we’re diving deep into the core principles of human motor control—something that’s been a mystery for so long. We’re not just building models; we’re uncovering the fundamental mechanics of how the brain and muscles work together,” says Mathis.

Part-to-Whole Training: A key strategy

The success of the AI model lies in its learning methodology. Drawing inspiration from sports science, where complex skills are learned in parts before being executed as a whole, the AI model was trained using a curriculum-based approach.

The curriculum learning approach broke down complex hand movements into smaller, more manageable tasks. “To overcome the limitations of current machine learning models, we applied a method called curriculum learning. After 32 stages and nearly 400 hours of training, we successfully trained a neural network to accurately control a realistic model of the human hand,” says Alberto Chiappa, a member of the research team.

Role of motor primitives in AI and neuroscience

A key factor in the model’s effectiveness is its ability to recognize and leverage basic, repeatable movement patterns known as motor primitives. This method not only enhanced the AI’s ability to learn but also offered insights into how the human brain learns motor skills. The study’s findings could contribute to the development of neuroprosthetics that mimic human movements more closely.

Each task requires a specific set of motor primitives to be performed effectively, highlighting the importance of understanding and replicating these movement patterns in AI models and prosthetics.

Conclusion: AI's role in biomedical advancements

The combination of AI and biomechanics is offering new possibilities in understanding the intricacies of human movement. With further research and development, these insights can pave the way for more advanced neuroprosthetics, rehabilitation tools, and improved motor control for patients with mobility impairments.

MEDICA-tradefair.com; Source: École Polytechnique Fédérale de Lausanne (EPFL)

More about EPFL: www.epfl.ch/en