Advances in human prostheses advance at high speed thanks to research in robotics. Although there are many projects to build functional hands, the complexity of this limb is a challenge for researchers.
Now, a group of Japanese scientists presents a control system for a prosthetic hand that requires very little training and is very accurate.
As published by Science Robotics, a team of researchers from the University of Hiroshima (Japan) and the Hyogo Institute of Support Technologies tested its new myoelectric hand – which uses electrodes to detect the tiny electrical signals that occur when a muscle in the body it contracts or flexes- in an experiment with seven participants, one of them with amputated upper limbs.
Akira Furui (principal investigator of the study) and his colleagues created a 3D printed prosthetic hand with five independently driven fingers and developed a new control system based on muscle synergy, that is, when several muscles combine to perform a concrete movement.
Thus, the combined activation of a particular muscle group drives the movement of a finger to, for example, point to the index. Other movement is achieved with different combined movements of muscles and their transitions.
The article shows how participants in the experiment who did not have amputated limbs performed 10 different movements of the fingers with more than 90% accuracy, and the prosthesis needed very little training: only five movements per finger.
For his part, the amputee participant – who carried a personalized myoelectric hand – performed simple and combined movements with approximately 92% accuracy.
In the future, researchers intend to further facilitate the use of the system, especially focused on relieving muscle fatigue after prolonged use of the prosthesis.
According to the note that accompanies this study, the myoelectric prostheses available in the market have a high cost and require a long period of training to be able to perform a variety of movements.
Advances in robotic prostheses
In recent years, many projects have emerged to adapt robotics to human prostheses, such as the Luke arm created by Deka for the military agency DARPA.
Open Bionics and its ‘Hero Arm’, printed in 3D, open source and tested in clinical trials with children in the United Kingdom, sells its myoelectric prosthesis for about 5,000 euros and has a store with molonas housings to give the prosthesis an appearance of the uniform of a superhero.
A similar development is the one presented last year by YouBionic, whose bionic arm ‘Arm’ can also be printed in 3D to save costs and is compatible with other robotic arms, acting as an actuator, explained in Engadget.
Several robotic prostheses have shown extraordinary precision, without the need for adjustment buttons for various modes of movement. However, they require a long and personalized training.
In the robotic prosthesis developed by the Georgia Institute of Technology (USA) in 2017, the researchers used ultrasounds in an amputated patient to gather more data of their muscular movements; using ‘machine learning’ techniques (something that is applied to other robots controlled with brain waves, for example) managed to describe the movements of each finger, and even the intensity of each movement.
Applied this information to the driver of the prosthesis, the result was spectacular: his patient could play the piano.