Droplet morphology plays a crucial role in diverse applications, from targeted particle deposition to novel drug delivery systems. While traditional particle-based methods often rely on numerical simulations, real-world experimental validation and data acquisition are paramount for advancing our understanding of particle behavior. This presentation introduces DropTrack [1,2], a cutting-edge computer vision tool designed for precise, real-time tracking of both droplets and individual particles within fluid flows directly from video footage. DropTrack's robust architecture leverages the YOLO algorithm for efficient particle detection and the DeepSORT algorithm for high-speed, accurate trajectory extraction at over 30 frames per second. By providing detailed, high-resolution particle trajectories, DropTrack offers a powerful experimental complement to established numerical methodologies such as the Lattice-Boltzmann Method (LBM). This capability enables researchers to deduce particle-particle interactions and derive general equations of motion for particles within complex flow environments, significantly assisting in the fundamental understanding of particle transport phenomena and validating computational models. The second part of the talk delves into the application of advanced decision-making algorithms, particularly reinforcement learning, to automate droplet production and achieve fine control over experimental conditions. We will discuss how integrating DropTrack's real- time feedback with reinforcement learning enables the precise manipulation of particle-laden droplets. We will showcase this integrated approach in a novel targeted drug delivery application, specifically focusing on the encapsulation and protected transport of highly sensitive probiotic material for precise delivery to the lower abdomen. This demonstrates how advanced particle control, informed by real-time visual tracking, can address critical challenges in biomedical engineering.