RoboDrop: A Multi-Input Multi-Output Control System for On-Demand Manipulation of Microfluidic Droplets Based on Computer Vision Feedback

Active control of individual picoliter- to nanoliter-sized droplets in a network of microchannels is vital to make droplet microfluidic platform an enabling technology for single-cell or single-particle analysis, which has found application in areas such as advanced manufacturing, material synthesis, life science research, and personalized medicine. The challenge manifests from the coupled dynamics between droplet motions and network inlet pressures, which must be overcome in order to control individual droplets successfully. In this article, we proposed a generalized approach for modeling and controlling droplet position. The model is validated experimentally and used in a series of multi-input multi-output linear-quadratic regulation controllers. The controllers obtain feedback from computer vision and actuate electropneumatic transducers to yield desired droplet movements. The ability to dynamically generate, trap, merge, split, and sort droplets according to real-time user demand is demonstrated with successful experimental results.