This repository contains the instructions/code/files for the assembly and operations of a Pen Plotter Driven Liquid Handler and Pipette Driven Liquid Handler as well as a collection of hands-on tutorial notebooks. Both devices are designed as low-cost automation tools to produce self-driving labs (SDLs). The pen plotter-driven liquid handler is created using rigid pre-built spatial components that are modified to fit a user-developed fluidic system. The pipette-driven liquid handler is designed using modular components to provide the most flexibility, paired with a wireless autopipette.
The goal of the provided tutorial is to walk through the essentials of AL from start to finish, providing both conceptual explanations and hands-on code examples. A basic understanding of ML is assumed. If you're new to ML, we highly recommend reviewing Part 1 of our User’s Guide series. This guide is structured into three main tutorial notebooks, each building on the last for a smooth learning experience. Major points of interest from the guide provided are discussed below.
We begin with a simple, two-dimensional example of Bayesian Optimization (BO) applied to a black-box function. This notebook introduces the foundational concepts of active learning, inccluding data seeding, fitting surrogate models to data, and applying various acquisition functions to choose new sampling points.
Finally, we apply AL to a real experimental dataset involving enzymes. This notebook demonstrates how active learning can be used to efficiently select informative experiments and accelerate discovery in scientific research. This notebook explores: seed library generation (initial sampling) and closing the loop in the context of an SDL.
Experimentation is inherently difficult because most methods require substantial refinement, calibration, and validation before high-quality, reliable data can be collected. In most cases, experimental domains have multiple variables (i.e., high dimensionality), thus requiring their simultaneous optimization for single and multi-objective targets. Traditional experimental approaches rely on trial-and-error methods guided by rational decision making, but these become increasingly inefficient and ineffective as complex interactions between inputs limit our ability to capture underlying trends using conventional statistical approaches. Active learning and machine learning (AL/ML) combined with automation represents an indispensable approach for future laboratory productivity. However, a steep initial learning curve and high costs of instrumentation pose substantial barriers to adopt this powerful approach. To democratize access, we provide a comprehensive tutorial covering both the computational skills and hardware implementation necessary for self-driven experimental workflows. The accompanying open-source, low-cost liquid handling platforms offer practical templates for researchers adopting self-driving lab (SDL) methodologies.
Apostolos Maroulis, Dylan Waynor
Last Updated 10/02/2025