CeyeHao: AI-driven microfluidic flow programming with hierarchically assembled obstacles and receptive field–augmented neural network

Abstract

Microfluidic fabrication technologies are increasingly used to produce functional anisotropic microstructures for broad applications. However, the limited flow manipulation methods hinder the production of intricate microstructure morphologies. In this work, we introduce CeyeHao, an artificial intelligence–driven flow programming methodology for designing microchannels to perform unprecedented flow manipulations. In CeyeHao, microchannels containing hierarchically assembled obstacles are constructed, offering more than double flow transformation modes and immense configurability compared to state-of-the-art methods. An AI model, CEyeNet, predicts the transformed flow profiles, reducing computation time by up to 2700 folds and achieving up to 97 and 90% accuracy with simulated and experiment results. CeyeHao facilitates microchannel design in both human-guided and automatic modes, enabling creation of flow morphologies with highly regulated geometries and elaborate artistic patterns, along with precise topology manipulation of multiple streams. The superior flow manipulation capability of CeyeHao can facilitate broad applications from complex microstructure fabrication to precise reaction control.