What you need to know
- PepsiCo and Microsoft worked together to improve the process of making Cheetos.
- AI is trained on parameters and then used to measure and adjust the machines that produce Cheetos.
- PepsiCo's trial runs have been successful, and the company plans to scale the systems across its production lines.
From their crunchy texture to their famous dust that gets on your fingertips, Cheetos are a popular and tasty snack. You might think that the oblong bumpy shapes of Cheetos are random, but they're actually the result of complex production and planning. A new blog post from Microsoft outlines how its Project Bonsai is used by PepsiCo to produce Cheetos. The AI that's powered by Project Bonsai has already been tested at a pilot plant and could soon be used in other areas of production.
PepsiCo and Microsoft worked together to develop an AI that can recommend adjustments to machines used to produce Cheetos. The system uses data from a computer vision system and has yielded results that PepsiCo is happy with.
"The project brought together a mix of technology, applied modeling skills, and subject matter expertise to create innovation on the factory floor," said Dylan Dias, CEO of Neal Analytics, which partnered with Microsoft and PepsiCo as part of the project.
Cheetos are made on a machine called an extruder. Usually, people would manually check Cheetos to see if any adjustments need to be made to the extruder. With Project Bonsai, the production line can be monitored "almost continuously," and can then recommend changes. An operator can then approve the changes, or the system can be set up to make changes automatically.
The initial results of pilot testing indicate that Project Bonsai does a good job adjusting the extruder to maintain consistency and quality.
The ultimate goal of this AI solution is to run autonomously. As outlined by Microsoft, "By empowering it to monitor the product and adjust the extruder line continuously and independently, the company expects to consistently maintain Cheetos quality and produce more throughput."
