- The Washington Times - Saturday, June 30, 2007

Today, the octofruitopus and the intersection of technology and politics, this time in the form of a robotic fruit picker being designed by Vision Robotics. If it works, and I can’t think of any compelling reason why it won’t, a lot of migrant farmworkers are going to be out of jobs. This presumably will affect the politics of immigration.

Although many see immigration in terms of nationalism or of left-right politics, for many businesses the availability of Mexican labor is a matter of commercial survival. Agribusiness and construction are examples.

In both, businessmen worry that restrictions on immigration may leave them catastrophically without labor. With perfect predictability, they begin looking for a technical solution — in the case of citrus growers, a computerized orange picker. Enter Vision Robotic of San Diego, founded in 1999.

The company figures that advancing technology and falling costs now make it possible for robots to do things that they couldn’t before. The firm is inventing an orange picker. This is more important than it sounds. It represents a further step in the inevitable advance of robots into the job market.

Picking fruit mechanically is not an easy thing to do.

First, the system has to find the oranges, which means it has to have vision. Solid-state cameras are both cheap and reliable these days, so this part isn’t too difficult. Seeing the oranges isn’t enough. The picker has to figure out where they are in three dimensions, which is lots harder than just seeing orange spots against a green background. Next it has to figure out what order to pick them in for best efficiency.

This has got to be a bear of a programming job.

Finally the picking arms — there will be eight of them, which is why I think the company should call it an octofruitopus — have to grab the oranges, preferably without destroying the tree or the orange.

This gets to be a whole lot of calculation.

Computational power has gotten so cheap, though, that what would have taken a pricey supercomputer not so long ago now can be done with commodity chips. Note that while an orange picker is highly specialized, the technology is not. Software that understands how to snatch oranges can be modified to do all sorts of things.

Now, long ago you could read books of science fiction that said one day machines would do all the work and humanity would starve in bulk. This was a bit overwrought. However, there has indeed been a steady erosion of jobs for the unskilled and poorly educated.

Backhoes replaced ditch diggers, answering machines have replaced a lot of unskilled secretaries, industrial robots have ousted lots of assembly-line people, and so on.

If a fruit picker can be made to work at a reasonable price, many thousands of jobs will disappear. While it is not politically correct to dwell on the point, most of these will be people who will have trouble finding other work. Many can’t read or speak English well.

They, too, need to eat.

The question becomes, “Where is all of this taking us?” What jobs cannot in principle be done by machines? The answer may be surprising. I have seen computer-driven cars that handle themselves pretty well in traffic, and software that does a remarkably good job of diagnosing disease. (You feed it blood pressure and test results, it asks about symptoms, and voila! You’ve got appendicitis.) Which brings us back to politics. Can the economy somehow generate work for those whose jobs are automated? For the reasonably educated, probably. For the utterly unskilled, I don’t see how, but maybe.