Autonomous Supervisory Control in Nuclear Power Plants: Fortier's Path to Autonomous Operations

To be a viable clean energy source, nuclear power must be competitively priced and economical to produce. Lauren Fortier, a second-year doctoral student in the Department of Nuclear Science and Engineering at MIT, is advancing this objective by developing remote operation protocols for autonomous control of nuclear plants. Her background as a naval nuclear operator, including supervision of plant operations on a U S aircraft carrier, gave her firsthand experience with the complete reliance on nuclear power and the intense demands of manual workflows. The question driving her work is not only how to automate, but how to automate without sacrificing safety, transparency, or trust. This article analyzes Fortier's approach to autonomous supervision, the control theory underpinning it, and the collaborations that enable practical progress toward scalable, safe nuclear automation.

Analytics — economic and technical viability of autonomous supervisory control in nuclear power plants

The economics of nuclear energy depend on balancing capital costs, operating costs, and capacity factors. Traditional plants justify large staffing and manual procedures because they operate at high capacity with stringent safety and regulatory demands. Fortier reframes the problem by asking how a supervisory control system can reduce labor intensity without eroding safety margins or regulatory compliance. The core premise is that a well designed autonomous supervisory layer can lower staffing requirements in remote or rural microreactors while preserving, or even improving, reliability and traceability of decisions. This shift hinges on three interdependent factors: control system transparency, robust human–machine interaction, and rigorous validation of automation under real and simulacrum conditions. In practical terms, the economics of autonomy emerge when the system frees scarce operator time for anomaly handling and optimization, rather than enforcing routine steps that humans could perform without error. This is where Fortier ties control theory to real-world plant dynamics, highlighting the need for predictable, auditable behavior rather than opaque, data-driven black boxes. The financial argument for autonomy rests on the alignment of event-driven automation with discrete, verifiable transitions in plant state, enabling safer operation with lean staffing and lower long-term operating costs. Fortier argues that the economic case strengthens as the automation framework scales to small, distributed reactors and as the integration with existing safety cases matures. The broader implication is that microreactors in rural grids can achieve competitive levelized costs if automation delivers consistent performance with transparent decision pathways and straightforward regulatory validation. The analytical picture, therefore, blends cost-performance trade-offs with the physics of reactor operation and the engineering of cyber-physical control loops. The result is a path to competitiveness that does not rely on unproven technology but on a principled, verifiable supervisory architecture with explicit human intervention points when needed.

From a technical perspective, autonomy in nuclear plants requires a reformulation of operations as objective-driven sequences rather than fixed procedures. Fortier emphasizes the economics of contextually appropriate automation: the system should generate a sequence of events that advances toward a stated objective, acknowledging that plant conditions vary and that procedures must adapt rather than rigidly repeat. This approach reduces brittleness and aligns with the realities of thermal hydraulics, plant transients, and safety interlocks. The analysis also considers the role of finite state automata in creating discrete, event-driven transitions that are easy to audit and validate. Unlike data-driven AI, which can be opaque and difficult to validate in regulatory terms, finite state methods yield transparent, traceable decisions. The upshot is a design philosophy that preserves operator confidence by providing clear, stepwise reasoning for automated actions and for the moments when human oversight must intervene. In short, the analytics point to a sustainable economic and safety envelope for autonomous supervisory control in nuclear power plants, provided the system remains interpretable, auditable, and aligned with rigorous plant physics and safety standards.

Contrast — humans and machines: strengths, limitations, and the path to trust

The legacy approach to nuclear plant operation centers on human-centric workflows. Operators wield deep expertise and situational awareness but contend with cognitive load, fatigue, and the risk of error in complex, high-stakes environments. Fortier foregrounds the strengths of human operators: nuanced judgment, rapid adaptation to novel anomalies, and the ability to interpret evolving conditions within a broader plant context. The counterweight is the volume and repetitiveness of routine procedures, which are ripe for automation without compromising safety. The autonomous supervisory model seeks a productive tag team where machines handle well-defined, repetitive, event-driven tasks, and humans intervene selectively when interpretation, ethical judgment, or regulatory nuance is required. The conflict is not about replacing humans but about preserving accountability and ensuring a meaningful role for operators in a transparent control loop. This distinction matters for acceptance and regulatory alignment because it reframes automation as a cooperative system rather than a wholesale substitution. The human factors dimension is essential: a trustworthy interface must communicate the system state, anticipated transitions, and the rationale for automated steps in a way that humans can audit, critique, and, if needed, override. In practice, that means designing a human–machine interface that supports strategic intervention rather than micro-management, and that provides operators with a clear mental model of the automation’s behavior under varying plant conditions. The contrast also highlights the importance of staged deployment. Fortier envisions a gradual ramp from manual procedures with automated guidance to closer-to-autonomy operation, with explicit safety valves and intervention thresholds to maintain trust and regulatory compliance. This progression helps bridge the gap between theoretical elegance and practical deployment by building confidence through incremental demonstrations of reliability and transparency.

Crucially, Fortier integrates human factors research into the engineering workflow. Her collaboration with the Idaho National Laboratory and its Human System Simulation Laboratory exposed her to methodologies for evaluating how operators interact with cyber-physical systems. The takeaway is not merely that humans can supervise automated actions but that the human role changes as automation advances. The human–machine symbiosis must be designed so that humans can anticipate automated moves, understand why a particular sequence emerged, and step in when the system’s assumptions no longer hold. This collaborative dimension is essential for the acceptance of autonomous nuclear plant operation because it anchors the technical design in the realities of human performance, cognitive workload, and safety culture. The contrast, therefore, reveals a path forward where autonomy reduces routine cognitive load while maintaining the centrality of human judgment in threshold moments that demand ethics, regulatory alignment, and professional accountability.

The practical implication is that autonomous supervision in nuclear power plants thrives on deliberate coupling of control theory with human factors. Fortier emphasizes transparent decision logic, not opaque optimization, as a foundation for trust. She also notes that regulators must see a clear evidentiary trail linking automated actions to plant safety and performance outcomes. In this sense, the contrast becomes a roadmap: automate what can be validated and routinely executed, keep humans in the loop for the interpretive and regulatory-critical steps, and maintain an interface that communicates both the system's state and the rationale for its actions. This nuanced stance is essential for the next generation of remotely operated, distributed reactors where the scale of deployment magnifies both the benefits and the risks of autonomy. The result is a practical balance between the efficiency gains of automation and the vigilance required to maintain public safety and regulatory trust.

Cause and effect — from finite state automata to safe, auditable plant behavior

The methodological core of Fortier's approach is a control framework built around objective-oriented operations and finite state automata. The core idea is to specify objectives and allow the supervisory controller to determine the sequence of events necessary to achieve those objectives under current conditions. This contrasts with rigid, procedure-driven automation that can fail when plant dynamics diverge from the expected baseline. The finite state automata approach yields discrete, event-driven transitions that are inherently auditable and interpretable, addressing a central concern in nuclear safety: how to justify each automated action to regulators, operators, and the public. The event-driven nature means the system reacts to conditions such as temperature shifts, pressure transients, or equipment status changes with well-defined if-then transitions, rather than drifting through a black-box optimization. This clarity is essential for validation, verification, and regulatory acceptance. It also provides a clear mechanism for human intervention when a state transition reaches a boundary where operator insight is necessary. Fortier's internship at INL deepened her appreciation for the need to verify that each state transition corresponds to verifiable physical changes, not merely symbolic promises. The consequence is a control architecture whose reliability can be demonstrated through discrete tests, simulations, and controlled demonstrations, which is precisely what regulators require for licensing new automation layers in nuclear environments.

The cause-and-effect chain in this framework links sensing, decision logic, and actuation in a transparent loop. Sensing detects a plant condition, the supervisory controller selects a safe, objective-driven sequence of actions, and actuators implement those actions while logging the rationale for traceability. Because the framework operates on discrete events, the system can be stress-tested across a broad spectrum of scenarios, from normal operation to partially degraded states. The advantages include predictable behavior, easier fault isolation, and straightforward backtracking to determine why a particular action occurred. Fortier emphasizes that this approach is not anti-innovation; rather, it provides a robust platform for future enhancements. When new safety features or regulatory requirements arise, they can be integrated as additional state transitions or decision rules rather than as monolithic, opaque modifications. The result is a scalable model for autonomous supervision that remains aligned with plant physics, safety constraints, and regulatory expectations—a critical combination for realizing autonomous nuclear plant operation at scale.

Expert reconstruction — Fortier’s integrative path to deployment and scale

Fortier frames a stepwise progression toward autonomy that starts with guided operations and steadily adds autonomous capability. The first layer embeds automated procedures that walk operators through familiar tasks, reinforcing trust by showing that the machine executes steps the operator would naturally perform. As confidence grows, the system shifts toward objective-oriented automation, where a controller proposes a sequence of events to reach a defined plant objective, with human oversight as a final checkpoint. This staged approach reduces the risk of abrupt transitions that could erode trust or violate regulatory expectations. A central theme is the integration of a central supervisory control system rather than a patchwork of interlinked subsystems. The latter can become fragile, especially when human operators and automated agents must coordinate under diverse operating conditions. Fortier argues for a unified interface that presents a coherent picture of plant state, goals, and the automation's reasoning in a way that humans can audit, critique, and intervene when necessary. Her work emphasizes the necessity of explicit, auditable rationales for automated actions, which is essential for regulatory accountability and public trust. Through this lens, autonomous nuclear plant operation becomes less about replacing humans and more about designing a robust partnership that leverages machine precision and human judgment where each is strongest.

Collaboration plays a pivotal role in this reconstruction. Fortier works with a multidisciplinary team under Sacit Cetiner at MIT NSE, with joint appointments and joint research with Idaho National Laboratory, and with the Human System Simulation Laboratory at INL. These collaborations provide access to sophisticated human factors analysis, cyber-physical system design principles, and validated testing environments for control software under realistic plant dynamics. Industry engagement with Westinghouse during a 2025 internship provides practical testing of ideas on next-gen reactor designs, ensuring that the proposed supervisory architecture aligns with vendor capabilities and real-world constraints. The result is a research program that moves beyond theoretical propositions to demonstrations with tangible navigation paths for deployment across next-generation equipment. The Innovations in Nuclear Energy Research and Development Student Competition recognition in 2025 signals that the approach has captured attention beyond the lab, indicating that a credible path to commercialization exists when technical rigor is matched with stakeholder engagement and governance considerations. The expert reconstruction thus presents a credible blueprint: integrate transparent, event-driven autonomy with a clear human role, validate the framework through cross-institution collaborations, and chart a gradual, auditable path to deployment that scales to microreactors and distributed networks.

Ultimately, the Fortier model argues for a future in which autonomous supervisory control in nuclear power plants supports safer operations, reduces staffing burdens in remote settings, and accelerates the deployment of microreactors that can contribute to resilient energy systems. The work shows that the most compelling path to deployment combines control theory with human factors science, rigorous validation, and practical partnerships across academia, national laboratories, and industry. The result is not a speculative technology push but a disciplined program that aligns technical feasibility with regulatory accountability and market realities. The story of autonomous nuclear plant operation, as Fortier charts it, is a story of meticulous integration: a governance-friendly, transparent, and scalable framework designed to earn trust from operators, regulators, and the public.

In the near term, the supervisory control framework will likely prove most impactful in next-generation equipment and distributed microreactors where the calculus of staffing costs is most sensitive to automation. The synergy with control theory, finite state automata, and human factors research offers a robust foundation for scaling. Fortier remains optimistic about the road ahead, noting that the most valuable insights come from working with stakeholders outside her immediate bubble. The path to deployment requires not only technical breakthroughs but also a sustained orchestra of collaborations, standards development, and iterative demonstrations that build public and regulatory confidence in autonomous nuclear plant operation.

As the field moves forward, the central question will be how to maintain trust while expanding autonomy. Fortier’s work suggests a rigorous answer: preserve a central supervisory framework, provide transparent state transitions, and enable strategic human intervention where judgment matters most. The result is a practical blueprint for autonomous supervision in nuclear power plants that respects physics, safety, and the realities of energy markets. If this balance holds, autonomous nuclear plant operation could become a mainstream capability that enhances safety, reduces costs, and accelerates the adoption of distributed, clean energy systems.

In closing, the relevance of Fortier’s research lies not only in its technical merits but in its collaborative infrastructure. The alliances with MIT NSE, INL, Westinghouse, and industry partners demonstrate that real-world impact emerges when engineering rigor meets human-centered design and stakeholder engagement. The lesson for the field is clear: the most durable progress in autonomous nuclear plant operation will come from systems that are transparent, auditable, and co-created with those who will operate, regulate, and rely on them. The result is a credible, scalable path toward autonomous supervisory control in nuclear power plants that aligns safety, economics, and practical deployment goals.

Key takeaways

• Autonomous supervisory control in nuclear power plants can reduce manual workloads while preserving safety and regulatory alignment.
• The finite state automata framework provides transparent, event-driven logic suitable for audit and verification.
• Successful deployment requires human factors integration and cross-disciplinary collaborations with national labs and industry partners.
• A staged path from guided automation to objective-driven autonomy helps build trust and meet regulatory expectations.

Note on terminology and scope: the discussion centers on supervisory control systems that guide, monitor, and intervene in nuclear plant operations rather than generic industrial automation. The emphasis on auditability, transparency, and human oversight reflects regulatory realities and the need for credible deployment in sensitive energy infrastructure.

Validation, Verification, and Deployment Governance

To move autonomous supervision from concept to licensable operation, a formal validation and verification (V&V) framework is essential. This closes a critical gap by defining objective criteria, auditable evidence, and safety arguments regulators can review alongside the plant safety case. A governance model that weaves control theory, human factors, and regulatory expectations ensures the system behaves predictably under real plant dynamics.

• Define acceptance criteria that tie automation actions to measurable safety margins.
• Ensure each automated transition has verifiable physical underpinnings and logging to support audits.
• Preserve operator oversight with clearly defined intervention thresholds and rollback procedures.
• Engage regulators, utilities, and vendors early to align on documentation and standards.

Table: Validation and Testing Phases

These activities build a transparent link between sensing, decision logic, and actuation, enabling regulators to trace every automated action to a physical condition and a safety rationale.

Table: Deployment Timeline

With such a framework, autonomous supervision can progress in a controlled, auditable fashion, aligning technical merit with regulatory expectations and market realities.