The full system integration showing the combination of HD-sEMG sensors on a soft socket, soft multi-synergistic hand, and the real-time neural synergistic control. Credit: Patricia Capsi-Morales.
Recent technological advances have opened new possibilities for the development of assistive and medical tools, including prosthetic limbs. While these limbs used to be hard objects with the same shape as limbs, prosthetics are now softer and look more realistic, with some also integrating robotic components that considerably broaden their functions.
Despite these developments, most commercially available robotic limbs cannot be easily and intuitively controlled by users. This significantly limits their effectiveness and the extent to which they can improve people's quality of life.
Researchers at the Italian Institute of Technology (IIT) and Imperial College London recently developed a new soft prosthetic hand that could be easier for users to control. This system, presented in a Science Robotics paper, leverages a new control approach that integrates the coordination patterns of multiple fingers (i.e., postural synergies) with the decoding of the activity of motoneurons in people's spinal column.
"Our recent paper is an outcome of an ERC Synergy grant, for which my group at IIT and the group of Dario Farina at Imperial College London cooperated to establish a new, more 'natural' connection between users and their artificial hands by making human and artificial intelligence talk through a neural connection," Antonio Bicchi, co-author of the paper, told Tech Xplore.
Credit: Science Robotics (2025). DOI: 10.1126/scirobotics.ado9509
When humans complete a manual task, they typically coordinate the movements of their fingers in dynamic ways. These basic coordination patterns are known as postural synergies.
Bicchi, Farina and their collaborators devised a method to control prosthetic limbs that integrates postural synergies with the neural decoding of spinal motoneuron activities. In other words, finger coordination patterns are combined with the analysis of electrical signals originating from people's nervous systems, which can be used to predict the movements they would like the robotic hand to perform at a given time.
"The hand is built with a mix of soft materials, for the skin, the tendons and the ligaments, and of rigid materials, for the bones," explained Bicchi.
"The artificial 'bones' roll on top of each other, instead of turning around pins such as robot hands normally do. The tendon arrangement is such that the hand can adapt to the shape of objects to be grasped—thus reproducing an autonomous, intelligent grasping behavior observed in human hands."
An individual with a limb deficiency wearing the system during familiarization, and the other shows a healthy individual using the prosthesis as a bypass (below his arm). The study demonstrates the alignment between hand intentions and the prosthetic finger movements. Credit: Patricia Capsi-Morales.
A unique feature of the soft prosthetic hand developed by this team of researchers is that it does not only allow users to intuitively grasp objects, but also to manipulate them in specific ways. For example, it could allow users not only to grasp a bottle of water, but also to turn its cap between the robotic hand's fingers to open it.
Bicchi, Farina and their colleagues assessed the performance of the hand in a series of initial experiments, involving both people who possess all their limbs and those who require a prosthetic hand. Their findings were highly promising, as the hand was found to allow users to perform complex hand movements, manipulating objects more precisely and naturally than they would using other prosthetic hands.
In the future, the control approach and design devised by these researchers could be applied to the development of other prosthetic limbs. The soft prosthetic hand they introduced could also be improved further and tested in clinical settings, which could eventually contribute to its commercialization.
A video showing MN synergy-based control of the SoftHand Pro-2 by a participant without physical impairment. The participant explores the grasping and in-hand manipulation capabilities with various objects. This video shows the participant explore the control space without grasping any objects and perform examples of functional grasping and in-hand manipulation of various objects. Credit: Science Robotics (2025). DOI: 10.1126/scirobotics.ado9509
"We showed that there is a direct connection between two of the levels at which an intelligent organization in 'synergies' of the daunting multitude and complexity of human hands had been previously observed," said Bicchi.
"On one side, the phenomenology of human grasping was known to exhibit a latent space of a few patterns which dominate the daily use of hands. On the other side, the reconstruction of motoneuron signals used to command the hand from the spinal cord also showed a lower-dimensional space reducing their very large dimensionality. This paper bridges these two levels and demonstrates the efficacy of using the motoneuron language to more naturally control bio-inspired soft hands."
More information: Patricia Capsi-Morales et al, Merging motoneuron and postural synergies in prosthetic hand design for natural bionic interfacing, Science Robotics (2025). DOI: 10.1126/scirobotics.ado9509.
Journal information: Science Robotics
© 2025 Science X Network
