Posters at this year's conference

This research presents an autonomous synthesis platform using mobile robots to operate standard equipment and integrate multiple analytical techniques (LC-MS and NMR), enabling more robust, human-like decision-making for accelerating discovery compared to typical single-measurement systems.

We ran a remote workshop where students in Mexico learned how to use artificial intelligence to control real scientific experiments happening in Canada. By combining hands-on learning with live robot control over the internet, the program gave students a fun and practical introduction to the future of science and technology.

We introduce PurPOSE, an autonomous platform that can rapidly determine the solubility of multicomponent solids across extensive solvent–temperature spaces and select the most appropriate physical or empirical model for prediction. By producing comprehensive solubility landscapes with minimal manual intervention, PurPOSE accelerates purification workflow design and other data‑driven process decisions.

Can a $4 device make automated experiments tick with guaranteed precision? And can an idea in programming languages transform this hardware capability into an accessible tool for everyone to tinker with?

We have developed an automated flow system for polymer synthesis that can be controlled remotely and continuously monitored in real time using in-line NMR, IR, and UV-Vis spectroscopy. This setup enables the generation of large datasets for machine learning, accelerating the discovery of new polymer materials.

We deployed and evaluated a teleoperation system that allows remote human intervention in self-driving laboratories to recover from unexpected errors and maintain smooth operation. By using a 3D mouse to control a robotic arm, our approach improves precision, reduces recovery time, and enhances the resilience of automated experimental workflows.

We describe a new autonomous discovery collaborative facility at Argonne National Laboratory that involve a gantry line with a mobile arm connecting 5 robots and 2 characterization tools, and our first workflows that include an energy/water nexus project on exploration of membrane permeability to discover thin, rigid membranes, and highly selective membranes for water purification.

The objective of this study is to use a Deep Convolutional Neural Network (CNN) to denoise computed tomography (CT) scan images in order to segment crack structures and, subsequently, gain a better understanding of 2D and 3D crack propagation in various cementitious-based materials.

In this study, we are building an ML model to predict the solubility of select agrochemicals in solvent mixtures and concurrently developing a pipeline for automated solubility measurements of agrochemicals to build a database for experimentally measured agrochemical solubility.

Catalyst dissolution and surface restructuring pose significant challenges in electrocatalysis, often complicating the understanding of degradation mechanisms and stability-performance trade-offs. This poster emphasizes how physics-informed, data-driven modeling—spanning first-principles methods to machine learning—can unravel atomistic processes and guide the rational design of durable electrocatalysts.

We evaluated the ability of fine-tuned large language models (LLMs) to predict the synthesizability of hypothetical inorganic crystal structures. We also explored the explainability of these predictions by prompting the fine-tuned LLMs for reasoning, revealing various underlying factors that contribute to inorganic synthesizability.

Enabling robots to perform different placement tasks in the context of self-driving labs (SDLs) remains inherently challenging due to the wide variety of object geometries and placement configurations. To address this, we propose AnyPlace, a two-stage method trained entirely on synthetic data, capable of predicting a wide range of feasible placement poses for real-world tasks.

Computer vision is used to monitor and control the growth of crystals in solution, producing higher quality crystals and large datasets for further optimization.

We present MOF-ChemUnity, a unified knowledge graph that connects literature, crystal structures, and computational datasets to unlock the full scope of materials knowledge for MOF design.

We built a suite of tools to aid CMC (chemistry manufacturing and control) optimization of synthesis and formulation routes for potential therapeutics. Included we demonstrate a method for Bayesian optimization of chemical reactions and a solubility prediction platform.

Creating an end-to-end workflow to accelerate the manufacture and testing of directly compressed tablets

A novel aspect of this project is the integration of cyber-physical systems (CPS) to support real-time data acquisition and the development of predictive models for process selection.

We present the development and implementation of a modular robotic platform for the layer-by-layer synthesis of MOF thin films. These MOF films are studied for their electrocatalytic relevance, and this platform is enabled by a python-based orchestrator that provides a simple, user friendly interface."

To support autonomous materials exploration, we have developed an open-source middleware platform called NIMO (https://github.com/NIMS-DA/nimo). NIMO comes with multiple built-in AI algorithms to accommodate the diverse needs of materials science, enabling a wide range of automated and autonomous development workflows.

To address the challenge of optimizing organic molecules for semiconductors, a reinforcement learning model called RingQ-net strategically generates π-conjugated molecules by linking or fusing predefined ring sets. Our approach efficiently uncovers low-density but high-performing candidates more efficiently than graph- and text-based baselines in both computational cost and multi-property accuracy.

Automated blade coating workflow integrated with custom curing and surface characterization modules within a liquid handler robot.