Large-format additive manufacturing depends on a narrow thermal window: When a new layer is deposited, the layer below must be hot enough to bond but cool enough to hold its shape, a balance that traditionally required constant human supervision. Researchers at the U.S. Department of Energy's Oak Ridge National Laboratory built on an existing model to improve fault detection, adding a controller that monitors layer temperature for a variety of machines and materials through low-cost thermal cameras mounted at the machine's nozzle. Computer vision identifies deviations and adjusts print speed automatically. In a full-scale test, the system detected a 30% temperature shortfall and corrected it without human intervention.