Researchers from Santa Clara University, the New Jersey Institute of Technology and the University of Hong Kong have been able to successfully teach microrobots to swim using deep reinforcement learning, marking a substantial leap in the progression of microswimming ability.
There has been great interest in developing artificial microswimmers that can navigate the world in a similar way to naturally swimming microorganisms, such as bacteria. These microswimmers hold promise for a wide range of future biomedical applications, including targeted drug delivery and microsurgery. However, most artificial microswimmers to date can only perform relatively simple maneuvers with fixed locomotor steps.
In the researchers’ study published in Communications Physics, they reasoned that microswimmers could learn and adapt to changing conditions using AI. Just as humans learning to swim require reinforcement learning and feedback to stay afloat and propel themselves in various directions under changing conditions, so must microswimmers, albeit with their unique set of challenges imposed by physics on the microscopic world.
“Being able to swim at the microscale by itself is a difficult task,” said On Shun Pak, an associate professor of mechanical engineering at Santa Clara University. “When you want a microswimmer to perform more sophisticated maneuvers, the design of its locomotor steps can quickly become intractable.”
By combining artificial neural networks with reinforcement learning, the team successfully taught a simple microswimmer to swim and navigate in any arbitrary direction. When the swimmer moves in certain ways, they get feedback on how good that particular action is. The swimmer then progressively learns to swim based on their experiences interacting with the environment around them.
“Much like a human learning to swim, the microswimmer learns to move its ‘body parts’ — in this case three microparticles and extensible links — to propel itself and turn,” said Alan Tsang, assistant professor of mechanical engineering. at the University of Hong Kong. “It does this without relying on human knowledge, but only on a machine learning algorithm.”
The AI-powered swimmer is able to adaptively switch between different locomotor paths to navigate to any target location by itself.
As a demonstration of the swimmer’s powerful ability, the researchers showed that it could follow a complex path without being explicitly programmed. They also demonstrated the swimmer’s robust performance in navigation under perturbations arising from external fluid flows.
“This is our first step towards the challenge of developing microswimmers that can adapt like biological cells by navigating complex environments autonomously,” said Yuan-nan Young, a professor of mathematical sciences at the Institute New Jersey Tech.
These adaptive behaviors are crucial for future biomedical applications of artificial microswimmers in complex environments with unpredictable and uncontrolled environmental factors.
“This work is a key example of how the rapid development of artificial intelligence can be harnessed to address the unsolved challenges of locomotion problems in fluid dynamics,” said Arnold Mathijssen, an expert in microrobots and biophysics at the University of Pennsylvania, which did not participate. in the research. “The integration between machine learning and microswimmers in this work will generate further connections between these two highly active research areas.”
Story source:
Materials provided by New Jersey Institute of Technology. Note: Content can be edited for style and length.