🤖 Physical AI: Machine Intelligence and Atomic Material Simulation Combined! 🚀

[Hasan S. Cemkan]

Consider an autonomous robot transporting parts in a factory. Its software helps it avoid obstacles and find an efficient route. But how far it can travel between charges depends on the battery inside. Improving the battery requires understanding what happens at a scale no factory camera can see: examining the movement of ions within a material.

Physical AI brings these two engineering challenges together. At the machine level, AI helps equipment perceive and move through its environment. At the atomic level, AI-powered simulation allows researchers to examine the materials that influence what the equipment can do. Bringing both together offers opportunities to improve industrial performance long before a machine ever reaches the factory floor.

🔬 Valuing Experiments: Which Experiments Are Worth Doing?​

Materials researchers face more possibilities than they can test. Changing a composition or the arrangement of atoms can alter a material’s behavior, creating many candidates to examine. Every lab experiment consumes time and resources, so deciding what to do next is a critical engineering choice.

Computer simulation has long aided these decisions. Quantum mechanical methods can describe atomic interactions in detail, but the computational requirements limit the size of systems and the duration of behavior researchers can examine. A team may have to narrow its search before exploring most of the possibilities.

Machine learning interatomic potentials (MLIPs) make a broader search practical. These models learn from quantum mechanical calculations to predict energies and forces for atomic arrangements. Researchers can use these predictions to simulate atomic movement and interactions much faster than repeatedly performing the underlying calculations.

For a battery electrolyte, this could mean examining how easily lithium ions move within a candidate structure or formulation. The results can help researchers compare candidates and understand the mechanisms behind their behavior. They can direct lab work toward promising possibilities and investigate why others fall short.

🔍 Making Room for a Broader Search: The Example of Catalyst Development​

Catalyst development illustrates what this shift can mean at an industrial scale. The composition and atomic structure of a catalyst influence the reactions it can facilitate. Searching across these variables creates far more possibilities than a lab can reasonably synthesize and evaluate individually.

In a study announced in September 2026, ENEOS evaluated approximately 100 million candidate structures for oxygen evolution reaction catalysis, which is essential for hydrogen production from water electrolysis. AI-powered atomic simulation and accelerated computation helped identify prioritized candidates for synthesis and experimental validation. The announcement noted that a discovery process that traditionally takes years was reduced to a few months.

The immediate success was a computational search that generated candidates for physical testing. Determining their usefulness still requires experiments; this includes evaluating performance and durability under relevant conditions. Manufacturing feasibility and cost will also shape which candidates can progress toward industrial use.

For manufacturers, the value lies in how this changes the use of scarce experimental capacity. A team can explore a wider range of structures before choosing which ones to make. Lab time can then be spent examining the most promising candidates and resolving uncertainties revealed by the simulation. Faster computation improves the information available at the point a researcher decides on an experiment.

🎯 Starting with Performance Requirements: Areas of Application​

Manufacturers can use this approach by identifying a materials problem tied to a specific requirement. If a coating fails in a process environment, what mechanism needs to be examined? If battery performance limits a product, what material property could help address that constraint? These questions provide a basis for materials experts to decide whether atomic simulation can inform the next engineering decision.

Teams must also agree on the evidence needed to advance a candidate. A model useful for screening may need to be supported by additional quantum-mechanical calculations and targeted experiments before researchers can trust its predictions for a specific application. Operational engineers and materials experts should collaboratively define relevant conditions, keeping manufacturing requirements in mind throughout the search.

A requirement for longer run time for the mobile robot could lead to changes in navigation software, equipment design, or battery materials. Simulation helps materials researchers evaluate their options before committing lab time. Involving these researchers early in the engineering discussion gives them a clear performance target and provides designers with better evidence of what materials can help them achieve that goal.