Artificial intelligence (AI) can help improve the exoskeleton’s movements and make it easier for people wearing it to use less energy and reduce stress levels. A study published today in the journal Nature features a super-intelligent controller that uses AI data and virtual simulations to train robotic structures without the need for human testing. These devices improve the movement of people with or without…
Please subscribe to continue reading
Read without limits
Artificial intelligence (AI) can help improve the exoskeleton’s movements and make it easier for people wearing it to use less energy and reduce stress levels. A study published today in the journal Nature features a super-intelligent controller that uses AI data and virtual simulations to train robotic structures without the need for human testing. These devices improve mobility for people with and without disabilities, but fitting them requires work and money. “Human testing during the development of exoskeleton controllers takes a long time,” explains Hao Su, professor of mechanical and aerospace engineering at the University of North Carolina, lead author of the study. “Performance in a simulation environment improves the development process, and with a sufficiently accurate model, subjects with specific movement limitations or conditions can be imitated. It reduces the effort and stress of these subjects, as well as the logistics of collecting them for testing,” he adds.
The wearable exoskeletons in the study correspond to different body joints: hip, knee, shoulder and hand. These devices help to reduce the physical effort that a person has to exert, for example in the case of a factory worker or an astronaut. They make walking and squatting easier, while other walking aids only help. If the journey requires 100% effort, Su explains, the robot provides 30% assistance, so humans only require 70% effort. In the case of astronauts, the robot helps them avoid bone loss and thus imitate normal walking in places with microgravity: “Relieving the skeleton during a prolonged stay in reduced gravity increases the risk of loss of bone density and resistance”.
The device can be placed in one or more joints of the leg and has built-in motors to bend or extend the joint, Alexandra S. Voloshina, a professor in the Department of Mechanical and Aerospace Engineering at the University of California, explains in a paper accompanying the study. In this way, “it allows the leg muscles to relax and the system to take over the workload,” he adds. For Juan Ernesto Solanes, professor of systems engineering and automation at the Universitat Politècnica de València, who was not involved in the study, the system also offers other benefits: “It can be configured to transfer the weight of the arms to the core of the body, reducing physical stress. Its ability to reduce the physical exertion inherent in some jobs reduces the risk of musculoskeletal injuries in workers.”
Until now, the development of an exoskeleton required a long process of experimentation with many people. The study model is based on real data of a specific person, i.e. a digital twin of a person, to assist his movement. Artificial intelligence combines hip joint data (angle and velocity) with, for example, a profile generated by the exoskeleton. In addition, the cost of an exoskeleton is very high because it involves hours of human testing. In general, the price ranges between 46,000 and 116,000 euros. However, the studio device would be cheaper: “We plan to put it on sale at a price of about 2,000 and 3,000 dollars (about 1,800-2,800 euros) thanks to the innovation Hardware and on algorithms based on artificial intelligence,” adds Su.
One of the limitations of current exoskeletons is that they depend on an external power source to function, Solanes says. However, the research’s robotic hip structure generated the largest reduction in metabolic rate to date, with a reduction in energy expenditure of 24.3% when walking, 13.1% when running and 15.4% when climbing stairs. “This work makes science fiction a reality and allows people to spend less energy performing various tasks,” adds Su.
Exoskeletons often don’t take into account the user’s comfort or restrict their natural movement, as Solanes explains: “The human-machine interaction needs to be as natural as possible to make the device useful and less disruptive.” He adds: “It can have the opposite effect than it is intended to.” desirable, cause mental fatigue or physical injury to the user. “Mental fatigue is caused by having to constantly adapt to the assistance provided by the exoskeleton, which prevents natural interaction.”
The system’s AI-driven algorithm helps both young and old people. The hip exoskeleton offers improvements for able-bodied people in activities such as walking, running and stair climbing. “This is reflected in a reduction in the metabolic cost of performing these activities, allowing the user to perform them for longer periods of time,” explains Su, who adds that they have also developed a pediatric knee joint exoskeleton that improves mobility. children with cerebral palsy in a community setting. Cerebral palsy affects almost one in 500 born in Spain, according to the Spanish Confederation of Associations for the Care of People with Cerebral Palsy.
Outside research experts agree that the method is valuable for research on exoskeleton development. But Massimo Cenciarini, professor of mechanical engineering at the Universitat Politècnica de Catalunya, questions whether it would work for activities like jumping or more dynamic tasks like turning, standing up or sitting down. Despite the great progress in controlling the exoskeleton, Cenciarini admits that it may require more work in patients with partial spinal cord injuries or stroke patients.
The system can generate errors if the training data is skewed or incomplete, or due to bad architecture choices, Su admits, and if the task is very difficult, it will struggle to provide a good solution. Despite this, Cenciarini predicts: “This method could lead to speeding up the development of control policies for other exoskeletons and accelerate the development of exoskeletons for assistance and rehabilitation in neurological disorders that affect movement, such as Parkinson’s disease or stroke.”
You can follow EL PAÍS technology in Facebook y X or subscribe to ours here newsletter.
