Toyota, Invisible AI team up to deploy an AI computer vision system in factories

The news: Toyota is expanding a collaboration with Invisible AI with a goal of making its North America factories safer and more productive for higher-quality manufacturing.

  • The latest angle of the two-year partnership involves the rollout of the Austin, Texas-based startup’s AI computer vision platform in 14 of Toyota’s manufacturing centers, per Forbes.
  • Invisible AI, which was founded in 2018, will deploy its 500 edge AI devices at each plant. The devices feature a built-in Nvidia Jetson chipset, 1 TB of storage, and a high-resolution 3D camera to track the entire factory floor, per Forbes.
  • In addition to monitoring overall operations, the system can analyze human bodies in motion, including movement of the torso, limbs, and joints, looking for red flags, per Core77.

Better cars, faster: Automakers are steadily embracing automation and AI in manufacturing with the hopes of continuous operational improvement.

  • Although security cameras are commonplace in industrial settings, Invisible AI’s system represents a deeper, more analytic level of surveillance.
  • The AI computer vision model reportedly constantly processes data from video feeds, yielding real-time data and insights about what could be improved for better operations.

Self-conscious factories: Generally speaking, the more information and insights a company has, the better. As Invisible AI CEO Eric Danziger told Forbes, “If you can’t see problems, you can’t solve them.”

  • And yet, while AI surveillance systems would work well for a workforce of robots, human awareness that a computer is constantly watching and analyzing their movements could have a mental toll and alter behavior in ways not fully understood.
  • While the movement analysis could help workers improve ergonomics and prevent repetitive motion injuries, such scrutiny could also risk mobility discrimination.
  • Toyota and Invisible AI have said that the data will be anonymized outside of immediate production levels, including by blurring faces.
  • However, facial recognition is only one identifier. Individuals could also easily be recognized by other physical attributes, such as a limp.

The big takeaway: The technology can deliver unprecedented insights that could help eliminate inefficiencies and safety hazards in the automaking process. However, it raises a question: Would corporate executives welcome the same level of dissection of their work that their blue collar colleagues will receive?

"Behind the Numbers" Podcast