Ford's AI cameras are catching hard-to-spot manufacturing errors. It could solve the company's costly recalls.

Business Insider

4 hours ago

Every year, more than 300,000 F-150 pickups roll off the line at Ford's Dearborn Truck Plant. The vehicle — America's best-selling car — has a complex lineup of trim levels, electrical hardware, wiring harnesses, badges, fenders, and seat variations.
As machines whir and workers assemble components with gloved hands, certain new members of the company's quality-assurance team never get distracted: AI-powered cameras.
Ford has rolled out two in-house artificial intelligence systems, called AiTriz and MAIVS, to detect factory defects in real time. AiTriz, implemented in December 2024 and named after its Spain-based creator Beatriz Garcia Collado, uses machine learning and video streaming to catch millimeter-scale misalignments. MAIVS, which debuted in January 2024, relies on still images from smartphones mounted on 3D-printed stands that verify all of the correct parts are properly mounted on vehicles.
Together, they're helping workers pinpoint issues on the spot — before problems turn into costly warranty claims, recalls, or rework. It's an area Ford is hoping to improve upon after leading the automotive industry in recalls in four of the past five years.
"Imagine you're an operator, and you're getting a vehicle every minute. You have a number of different tasks you have to complete, and every vehicle is different," Jeff Tornabene, who manages vision applications at Ford's Manufacturing Technology and Development Center, told Business Insider. "When your control system knows the recipe, and your vision system can check it, there are some helpful suspenders there that make sure you're putting the right part on the right vehicle."
AI could be Ford's fix for a costly recall problem
These tools could relieve a billion-dollar headache for Ford. This year, the automaker faced a record 94 recalls, primarily affecting cars built before 2023. It's the most safety bulletins any major car brand has ever posted in an entire calendar year, and it's still early August. Chrysler has the second-highest number of recalls among major automakers, with 21.
Those bulletins are costing Ford millions of dollars. For example, a fuel leak recall for 694,271 units of the popular Bronco Sport and Escape SUV models will cost $570 million, according to the Detroit Free Press.
The company's AI-supported systems are already making a significant difference on the automaker's manufacturing floors, Patrick Frye, an engineering manager at the Dearborn plant, said.
"It absolutely has helped from an operational standpoint," Frye told Business Insider. "Instead of having to wait for vehicles to hit end-of-line or certain checkpoints, it allows for better in-station process control."
That kind of speed matters. Before systems like AiTriz, many electrical issues wouldn't surface until the final inspection, if at all, Frye said. Fixing those later often required ripping out carpets or removing seats. And as Ford vehicles gain more screens, sensors, and self-driving hardware, even a slightly loose connection can cause major headaches. For human workers, the tiny difference between a loose part and a good connection can be impossible to catch.
"With our handwear to protect our hands, or the general factory noise, you don't always hear the click, or you can't always feel the snap of two connectors going together," Frye said.
Brandon Tolsma, a vision engineer at Ford's MTDC, said AI's real-time results are often the difference between a fast fix and a carpet-ripping problem.
"As the vehicle goes through the assembly line, it gets harder and harder to access some of these components," Tolsma said. "I can't stress enough how the real-time results are key in saving us time."
The systems are now installed at dozens of stations across North America — AiTriz at 35 stations, and MAIVS at nearly 700. While MAIVS is limited to still images, AiTriz's live video feed offers more precision and adaptability, especially in cases of occlusion (like if a worker walks through a camera's frame or when a sheetmetal component blocks an electrical connection).
Previously, some faulty connectors still passed final inspection, since they made basic electrical contact, Tolsma said. But if the plugs weren't fully seated, they could later jostle loose in real-world driving. Ford believes AI cameras are catching those millimeter-off subtleties that factory workers — even with extensive experience — might miss, even if it's hidden behind sheet metal and carpeting.
"As the operators are trying to do their job quickly, it can be kind of hard to see, especially if they're some distance away from the electrical connector," Tolsma said. "But with vision, a high-powered camera can see that difference pretty clearly."
The start of a smarter assembly line
Analysts are cautiously optimistic. David Whiston, a Morningstar analyst who covers Ford, told BI that AI shows long-term promise, but it's still in its early stages.
"I do think AI done effectively could reduce recalls over the next decade and beyond," he said. "But for now, there's no guarantee it'll work. Anything that's more preventative in not only manufacturing but in design and engineering should, in theory, reduce warranty and recall costs."
For now, Tornabene says the vision tools are meant to assist the manufacturing workforce, not replace it.
As models grow more complex, AI systems like AiTriz and MAIVS have the potential to support plant operators who are implementing more advanced sensors, computer chips, and processors to aid Ford's self-driving ambitions, Tornabene told BI. They're also helping to cut through factory noise, ensuring that each car is built as intended.
"This isn't the endpoint for us," Tornabene said. "This is just the start of the plans that we have to make this technology a quality game-changer."