Active Pixel Sensors Came From NASA

Let’s start with this – did you know that one of the most important pieces of camera technology was developed at NASA’s Jet Propulsion Lab in 1993?

That’s right: it’s the active pixel sensor that’s now built into smartphone systems and led to massive changes in computer vision.

As I pointed out in another piece I did a couple of weeks ago, this was pioneered by Eric Fossum, a Dartmouth professor of engineering. Scientists were trying to miniaturize cameras for space exploration. However, the technology quickly became commercialized, and now it’s providing quite a bit of value in several fields, including agriculture.

Smarter John Deere Tractors

Take John Deere, where modern tractor designs feature an active vision technology that allows for things like self-driving sprayers and tractors, as well as automatic picking and harvesting. There’s also a grain monitor technology that looks at the inside of the combine tank, or the grain as it’s being poured into a silo, and assesses aspects of the crop.

The technology is described this way by Dan Miller, Senior Editor at Progressive Farmer, who covered the company’s accolades in 2020 for the camera tech:

“(It) helps farmers see inside the combine’s grain tank and observe tailings so they can monitor the condition of harvested grain, right down to individual kernels. This technology is fueled by proprietary algorithms and provides farmers with information to make critical decisions in the moment, and to gather data over time to inform future actions.”

So many other aspects of AI are knowledge-based or diagnostic, as in healthcare, or focused on non-physical and digital systems. But in agriculture, this computing power and breakthrough hardware is turned toward inspiring robotic designs.

To Agriculture and Beyond

Reading an article about this technology, I came across a startling list of professions that experts suggest will be filled by machine labor in just a few years.

A number of them have to do with agriculture; for example, there’s the fruit picker, delivery driver, and, improbably, shepherd.

Imagine the lone humanoid robot on the hill, carefully guarding a lack of sheep from predators. It doesn’t need a lunch break or a dental plan. It will carry out its job with the ultimate reliability of, well, a machine.

Here are the other jobs in the list:

Factory assembly worker

Retail worker

Cashier

Bank teller

Security guard

Long haul truck driver

First of all, this conjures up visions of completely empty storefronts across America. Banks with no human beings inside. A Gap store where robots carefully fold the merchandise and tally up customer purchases.

It also shows us to what extent some of the most common jobs in America are going to be outsourced to machines. Raising the question, what will humans do for work?

Absent a universal basic income, which would solve the problem, we’re going to have some serious challenges just administrating societies in which the ways that humans earned a livelihood are suddenly gone.

Anyway, all of it rests on innovations like this camera unveiled at Dartmouth in the 1990s. MIT’s museum has some interesting exhibits related to this kind of hardware evolution, and the community is spending some time and effort looking into the origin of our new technologies and how they came to be. That gives us a window back to the early days before the computers started “thinking” and everything changed.