TUM Develops AI-Powered Soft Robotic Glove Restoring Grip to Paralyzed Hands

Researchers at the Technical University of Munich (TUM) have created a soft robotic glove that uses artificial intelligence and pneumatic technology to restore gripping ability to individuals with hand paralysis. Published in Nature Machine Intelligence, the glove reads extremely weak electromyographic (EMG) signals from the forearm, which remain even in patients who have lost full motor control. Deep learning algorithms analyze these signals in real time to detect the user's intent to grasp or release objects, activating tiny pneumatic artificial muscles that move the fingers smoothly and precisely.

Unlike rigid external exoskeletons, the glove is made from soft, flexible materials that mimic human skin and muscle properties. Microscopic valves control air pressure with millisecond precision, enabling natural-feeling movement without stiffness. Weighing less than half a kilogram, the glove is customized for different hand sizes and levels of impairment.

Clinical trials led by Prof. Gordon Cheng and Dr. John Nesor demonstrated significant improvements in patients' ability to perform daily tasks, along with increased confidence and reduced frustration compared to existing solutions. Beyond immediate assistance, prolonged use of the glove may promote neurological recovery by reactivating dormant neural pathways and enhancing brain plasticity, with some patients showing improved independent movement without the glove.

Future plans include expanding trials to stroke survivors and patients with peripheral nerve injuries, as well as integrating sensory feedback to allow users to "feel" objects they hold. This innovation is part of TUM's eXprt initiative, which combines expertise in engineering, neuroscience, and medicine to revolutionize medical rehabilitation through AI and soft robotics.