AI and Exoskeleton Robots

Researchers at North Carolina State University in the U.S. recently announced in the international academic journal Nature that they have succeeded in developing exoskeleton without going through human-wear and walking experiments.

The exoskeleton robot, which means that the skeleton that supports the body like an insect is outside the body, is a device that is mounted on the human body to strengthen movement. It is called a "wearable robot" and is being developed to assist the elderly and the weak in walking and to strengthen the military's combat capabilities.

The result of this study is that the "Wearing Robot" learns movements through simulation and has walking abilities in a short period of time.

Previously, it took a long time for development as it was a method for users to secure motion data by wearing and moving them, and to analyze them and apply them to the control of "wearing robots." It also had the disadvantage of requiring extensive correction work when users change.

To solve this problem, the researchers trained a "wearing robot" with physical activity data obtained with motion capture (motion recognition) technology.

They have learned their movements through millions of virtual learning, which is almost impossible with the way humans wear and operate.

The AI-integrated "Wearing Robot" quickly detects the user's movement and implements the optimal motion accordingly.

The researchers said that those who wore the latest robot showed 24 percent higher energy efficiency on walking and 15 percent higher on stairs than the general public. In the case of running, the wearing robot is 13 percent more energy efficient, allowing them to run longer.

The researchers said, "We can develop 'wearing robots' by applying this principle not only to the lower body but also to more diverse body parts," adding, “It can be an opportunity for personalized motion assistance robots to become common.”

Some predict that when AI combines with "wearing robots," there will be a time when they become proficient in martial arts and sports without learning like the movie Matrix.

AI is on the verge of realization as it can learn even precise human movements through virtual learning. For example, according to a study published in the international journal Science Robotics in April by Google DeepMind, robots that have learned from virtual learning to dribbling, shooting, and defense have implemented soccer skills in real games.

With "deep reinforcement learning" that combines deep learning and reinforcement learning, robots acquire soccer skills as they learn improvements through trial and error. Applying simulation-learned techniques directly to actual movements is called “Zero-Shot Transfer.”

There is also a growing expectation that if AI, which has learned the movements of the master through virtual reinforcement learning, is applied to "wearing robots," anyone can show off their outstanding skills.

For example, if you wear an AI-specific "wear robot," you can dribble and shoot like Lionel Messi, or if you wear a wearable jacket specializing in performance, you can perform as much as a world-class pianist or guitarist.

Although it fell short of this, researchers at the Georgia Institute of Technology in the United States introduced a "wearing robot" that allows more colorful performances by installing additional robot arms on the arms of drummers.

In 2021, researchers at the University of Waterloo in Canada developed an AI-based wearable robot that can walk on its own while avoiding obstacles like an autonomous vehicle. This method supports autonomous walking so that users do not have to manually operate them. Medical wearable robots that enable sophisticated surgery are also being developed.

Key Takeaways: AI and Exoskeleton Robots

• Advancements in AI-powered exoskeletons: Researchers have developed exoskeletons that learn movement through virtual simulation, eliminating the need for human wear and walk testing.
• Increased efficiency and performance: AI-integrated exoskeletons enhance human performance, enabling users to walk, climb stairs, and run with increased energy efficiency.
• Zero-Shot Transfer: AI can transfer learned skills from virtual simulations directly to real-world movements, potentially allowing anyone to perform at a professional level.
• Applications beyond physical assistance: AI-powered exoskeletons have the potential to revolutionize various fields, including sports, music, and surgery.
• Future possibilities: The combination of AI and exoskeletons could lead to the development of robots capable of performing complex tasks, such as martial arts or playing instruments at a professional level.
• Autonomous functionality: AI is enabling exoskeletons to move autonomously, eliminating the need for manual operation and opening doors for greater independence.