Challenges and attempts to make intelligent microswimmers

Abstract

The study of microswimmers’ behavior, including their self-propulsion, interactions with the environment, and collective phenomena, has received significant attention over the past few decades due to its importance for various biological and medical applications. Microswimmers can easily access micro-fluidic channels and manipulate microscopic entities, enabling them to perform sophisticated tasks as untethered mobile microrobots inside the human body or microsize devices. Thanks to the advancements in micro/nano-technologies, a variety of synthetic and biohybrid microrobots have been designed and fabricated. Nevertheless, a key challenge arises: how to guide the microrobots to navigate through complex fluid environments and perform specific tasks. The model-free reinforcement learning (RL) technique appears to be a promising approach to address this problem. In this review article, we will first illustrate the complexities that microswimmers may face in realistic biological fluid environments. Subsequently, we will present recent experimental advancements in fabricating intelligent microswimmers using physical intelligence and biohybrid techniques. We then introduce several popular RL algorithms and summarize the recent progress for RL-powered microswimmers. Finally, the limitations and perspectives of the current studies in this field will be discussed.