Autonomous vehicle technology deep dive at John Deere and Zoox

Engineering & simulationSpark AI

John Deere uses Spark AI's system to have humans review images from the autonomous tractor when it stops for a potential obstacle, allowing the system to distinguish real obstacles from false positives like dust or shadows and continue driving without interrupting the farmer.

“What we do instead of of bugging the farmer is we send these images to this company that is now part of John Deere, called Spark AI, and they have a really good system where they have 2 humans that independently look at the image … if it's not, if it's a false positive … Then um, then it says, no, there's nothing there. You can continue driving…”

Engineering & simulation

John Deere's autonomous tractor uses machine learning to detect obstacles via cameras and stop conservatively when in doubt, triggering the Spark AI loop described above.

“when in doubt, what we have is our machine learning said to be super conservative. When it doubt, it stops, right? But not all the reasons it stops are really because there's something in front, a real obstacle.”

Engineering & simulation

John Deere acquired Blue River Technology to bring computer vision and machine learning expertise for selective spraying, which is now integrated into their autonomy and perception stacks.

“John Deere acquired Blue River, to bring that technology, the sea and spray technology, uh, into John Deere's portfolio, but also to bring the expertise of perception of computer vision machine learning into John Deere…”