Researchers Create an Exoskeleton for People with Mobility Issues that is Lightweight and Easy to Use, thus Positively Impacting the Healthcare Robotics Market
Posted On April 16, 2022
Robotic exoskeletons have the potential to play a significant role in assisting an ageing population. They are essentially suits that people can wear to help them exert strength when their elderly bodies are unable to do so. However, the development of exoskeletons has been impeded because they are often hefty and, if not well regulated, can act as hindrances rather than help. As a result, it is essential to build exoskeletons that are lightweight and capable of assisting the user's efforts without impeding them.
A research team has created an exoskeleton robot that could assist persons with mobility problems. The technology is highly beneficial for Healthcare Robotics Market as it allows the skeleton to estimate the user's intentions effectively. The system uses a combination of lightweight material engineering and artificial intelligence.
The current study has two major components. First, the researchers created a lightweight, carbon fibre-based exoskeleton for the lower torso linked to users' thighs and lower legs. The exoskeleton was designed with highly back-drivable actuators to allow users to move freely even while the actuators were not engaged. Notably, the research team turned to artificial intelligence to see if it could predict how the user would move.
They employed a method known as PU-learning (Positive and Unlabeled) to teach the exoskeleton to read the user's intents based on the user's muscle activity measurements. The PU-classification approach makes use of ambiguous data. This is done by combining positively labelled data that the computer knows is accurate with other unlabeled data that could be positive or negative. Thus, allowing artificial intelligence to learn from data that is not completely labelled.
The experiment went well. The results were better compared to conventional systems that usefully labelled data in situations where user behaviour other than the target sit-to-stand motion can occur. This indicates that the method could be expanded to other movements as well.
The main feature of the present research is that when directing a robot to support the human movement, it is vital to create it based on the idea that humans will behave in ways different than the learning data.