Researchers from the RIKEN Guardian Robot project and collaborators have combined lightweight materials engineering and artificial intelligence to create an exoskeleton robot that could help people in need of assistance. This new device offers technology that would allow the skeleton to effectively guess the user’s intentions.
Robotic exoskeletons would be a solution to an aging population. These are basically suits that people could wear that would allow compensate for the lack of strength.
However, a defect prevents their development, they are still generally heavy. Also, if not properly controlled, they may hinder rather than help. This is why it is important to develop exoskeletons that are both light and capable of assisting the efforts of the user without hindering them.
An exoskeleton combined with machine learning
Researchers have developed a lightweight carbon fiber lower body exoskeleton that attaches to users’ thighs and lower legs. It was built with highly reversible actuatorsso that it does not hinder users’ movements even when the actuators are not activated.
The team used artificial intelligence to check if it was able to predict how the user wanted to move. They have used the PU, or positive and unlabeled, learning method to teach the exoskeleton to correctly read the user’s intentions. To make this possible, it would be based on the measurements of the muscular activities of the latter.
This method allows the use of ambiguous data, by combining positively labeled data with other unlabeled data which can be positive or negative. This allows artificial intelligence to learn from data that is not all labeled.
For the experiment, the participants performed various movements starting in the same way: standing up, crossing their legs, leaning forward and repositioning themselves on a chair. The exoskeleton used machine learning to guess when they were actually trying to get up and then provided assistance for the movement.
According to Jun-ichiro Furukawa of the Guardian Robot Project, first author of the paper, the results were better than those of conventional systems. These use fully labeled data in situations where the user’s behavior may differ from the sit-stand target movement. This indicates that the method could be extended to other movements.
“The key to our research is that when controlling a robot to assist human movement, it is important to develop it with the assumption that humans will behave in ways that are not in training data. »
SOURCE: MIRA NEWS
We would love to thank the author of this post for this awesome content
A smart robotic exoskeleton to help people with reduced mobility
You can view our social media profiles here , as well as other pages on related topics here.https://www.ai-magazine.com/related-pages/