Scientists at Idaho National Laboratory have successfully demonstrated autonomous remote operation of a research reactor in Indiana using artificial intelligence and cloud computing infrastructure, marking a significant advance in next-generation nuclear technology management.
The experiment, completed in mid-July, involved researchers at INL working in concert with teams from the University of Illinois-Urbana-Champaign Grainger College of Engineering and Purdue University to remotely adjust the PUR-1 research reactor located in Indiana. The demonstration relied on a sophisticated digital control loop system equipped with reinforcement learning capabilities to automatically regulate reactor power without manual intervention at the facility itself.
Three-State Network Powers Real-Time Adjustments
The technical architecture spanned three geographic locations. INL’s high-performance computing systems in Idaho connected through Microsoft Azure cloud servers in Virginia to transmit real-time instructions to the PUR-1 reactor in Indiana. Rather than requiring operators on-site to physically manipulate control rods, researchers in Idaho fine-tuned reactor power remotely, using a digital twin model to simulate physical forces and calculate necessary adjustments with precision.
The reinforcement learning model proved essential to the system’s autonomy. By analyzing reactor behavior patterns and physics simulations, the AI could recommend and execute control rod movements that minimized power fluctuations and sustained steady operation—all without human intervention during the live test.
“If you can model a reactor with enough fidelity, you can eventually let AI safely assist in operating it,” said Chris Ritter, director of the Scientific Computing and AI Division at INL. The successful demonstration validated that principle in a real-world scenario.
Digital Twin Technology Builds on Prior Work
The PUR-1 digital twin that enabled this experiment was developed by a team led by Stylianos Chatzidakis, an assistant professor at Purdue’s School of Nuclear Engineering. INL and Idaho State University had previously demonstrated digital twin capabilities for reactor systems in 2023, but this marks the first successful remote autonomous operation across such distances using cloud infrastructure.
“The successful demonstration greatly expands the kinds of experiments and control system research we can perform at PUR-1,” Chatzidakis noted, underscoring the broader research implications beyond this single test.
Alignment with Federal AI-for-Nuclear Initiative
The demonstration advances the U.S. Department of Energy’s Genesis Mission, a coordinated effort to deploy artificial intelligence throughout the nuclear energy technology lifecycle—from initial design and regulatory licensing through manufacturing, construction, and operational phases. As advanced nuclear reactor designs continue to proliferate across the country, the ability to remotely monitor and adjust operations using AI could streamline staffing requirements and enhance safety margins.
INL has emerged as the epicenter of advanced reactor development in the Mountain West. The laboratory has hosted multiple milestone demonstrations in recent years, including Deployable Energy’s Unity Reactor, which reached a key milestone ahead of a presidential deadline, and a fourth advanced reactor that surpassed federal performance targets. These achievements reflect the region’s growing role in the nation’s nuclear innovation strategy.
The remote control capability also carries implications for nuclear waste management and decommissioning operations. As cleanup operations at East Idaho nuclear sites seek local support on related regulatory matters, the ability to operate facilities remotely with reduced on-site staffing could factor into future facility management decisions.
What Comes Next
The research team has not announced immediate plans for expanded demonstrations, but the successful proof-of-concept opens pathways for scaled deployment. Federal funding for AI-assisted nuclear operations is expected to remain robust, particularly as the Biden administration’s successor administration prioritizes energy independence and advanced manufacturing. Additional tests may focus on more complex reactor scenarios or extended operational periods to build confidence in autonomous systems before any deployment in commercial settings.