Ford uses AI cameras to prevent costly recalls by spotting factory errors

At Ford’s sprawling Dearborn Truck Plant, where over 300,000 F-150 pickups, America’s best-selling vehicle, roll off the assembly line annually, the complexity of modern automotive manufacturing presents a formidable challenge. Each truck boasts a intricate array of trim levels, electrical hardware, wiring harnesses, badges, fenders, and seat variations. Amidst the whirring machinery and the gloved hands of human workers, a new, tireless member of the quality-assurance team has emerged: artificial intelligence.

Ford has recently deployed two in-house AI systems, AiTriz and MAIVS, designed to detect factory defects in real time. AiTriz, introduced in December 2024 and named after its creator Beatriz Garcia Collado, leverages machine learning and live video streaming to identify even millimeter-scale misalignments. MAIVS, which debuted earlier in January 2024, employs still images captured by smartphones mounted on 3D-printed stands to verify that all correct parts are properly installed on vehicles. These systems work in concert, empowering workers to pinpoint issues immediately, preventing them from escalating into costly warranty claims, extensive reworks, or damaging recalls—an area where Ford has sought significant improvement, having led the automotive industry in recalls in four of the past five years.

The stakes are high for Ford, as recalls have become a multi-billion-dollar burden. This year alone, the automaker faced a record 94 recalls as of early August, predominantly affecting vehicles built before 2023. This marks the highest number of safety bulletins ever issued by a major car brand within a single calendar year, significantly surpassing competitors like Chrysler, which recorded 21. Such bulletins carry a hefty price tag; for instance, a fuel leak recall impacting nearly 700,000 units of the popular Bronco Sport and Escape SUV models is projected to cost $570 million.

According to Patrick Frye, an engineering manager at the Dearborn plant, Ford’s AI-supported systems are already making a substantial difference on the manufacturing floor. He notes that the technology has significantly improved operational efficiency by enabling “in-station process control,” meaning issues are caught and addressed precisely where they occur, rather than much later in the assembly process. This speed is critical. Previously, many electrical problems might not surface until the final inspection, or even later, often necessitating disruptive and expensive fixes such as ripping out vehicle carpets or removing seats. As modern Ford vehicles integrate more screens, sensors, and hardware for advanced features like self-driving, even a slightly loose connection can trigger major malfunctions. For human workers, distinguishing a fully seated connection from a loose one can be nearly impossible amidst factory noise and while wearing protective gloves.

Brandon Tolsma, a vision engineer at Ford’s Manufacturing Technology and Development Center, emphasizes that the real-time feedback provided by AI is often the crucial difference between a quick adjustment and a labor-intensive repair. He explains that as a vehicle progresses down the assembly line, accessing components becomes increasingly difficult. The AI systems are now deployed widely across North America, with AiTriz active at 35 stations and MAIVS at nearly 700. While MAIVS relies on still images, AiTriz’s live video feed offers superior precision and adaptability, even handling situations where a worker or another component temporarily obstructs the camera’s view. Previously, some faulty connectors might pass a basic electrical continuity test during final inspection, only to jostle loose during real-world driving. Ford believes its AI cameras are now adept at catching these minute, millimeter-off subtleties that human eyes, even those of experienced factory workers, might miss, particularly when components are hidden behind sheet metal or carpeting.

Industry analysts, such as Morningstar’s David Whiston, view Ford’s AI initiative with cautious optimism, acknowledging its long-term promise despite being in its early stages. Whiston suggests that effective AI implementation could reduce recalls over the next decade and beyond, emphasizing that any preventative measure in manufacturing, design, and engineering should theoretically lower warranty and recall costs. Ford maintains that these vision tools are designed to assist the manufacturing workforce, not to replace it. As vehicle models grow more complex, integrating advanced sensors, computer chips, and processors to support Ford’s self-driving ambitions, AI systems like AiTriz and MAIVS are poised to become indispensable allies for plant operators, ensuring that every vehicle is built precisely to specification. This, Ford suggests, is merely the beginning of a transformative shift in its quality control strategy.