Surprisingly well balanced article that highlights different viewpoints on the issue:
>> Members of the $7.5 billion annual Washington agriculture industry have long grappled with labor shortages
Human pickers are getting scarce [..] “Young people do not want to work in farms, and elderly pickers are slowly retiring.”
>> The work is hard and dangerous, and has long drawn Mexican workers to central Washington [..] He estimated half the state’s farmworkers are immigrants who are in the country illegally [..] many of them have settled in Washington and are productive members of the community
>> ..the industry is deeply interested in alternatives to human labor..
>> The robotic pickers don’t get tired and can work 24 hours a day.
>> “They are scared of losing their jobs to mechanization,” [..] “A robot is not going to rent a house, buy clothing for their kids, buy food in a grocery and reinvest that money in the local economy.”
>> “They are scared of losing their jobs to mechanization,” [..] “A robot is not going to rent a house, buy clothing for their kids, buy food in a grocery and reinvest that money in the local economy.”
A robot won't do those things, but in the long run the ex-farm workers will move to urban areas where they'll get jobs, rent houses, buy clothing for their kids...
That's what happened a century ago when the first wave of mechanization in the first world began a trend that reduced agricultural employment from >90% of all jobs to <5%.
Everything will be fine and the everyone's standard of living will rise...in the long run. In the meantime, let's try to avoid a populist revolution.
> but in the long run the ex-farm workers will move to urban areas where they'll get jobs, rent houses, buy clothing for their kids...
Let's take Washington state as an example. Where are the workers going to move? Seattle? Where rent is ridiculously high? What jobs will they get when more and more jobs these days require college degrees and manufacturing is going away?
The job and housing market today is distinctly different from 100 years ago—we place higher emphasis on schooling and training and many jobs that do not require some form of schooling don't pay enough to support a family (mostly because they're not designed to).
Please don't construe my comment as endorsing some sort of populist revolution. I just find it very hard to believe hordes of migrant workers (many of which are here illegally, don't speak much English, perhaps have little schooling, etc.) will just waltz on into cities and out of thin air procure jobs that pay enough to cover rent, food, and clothing—especially when they only reason they can support a family on an apple picker's salary of a couple hundred bucks a day is because they live in the cheapest part of the state.
I mean, if it was that easy why wouldn't they do it already? Cities offer vastly more resources for poor folk than central Washington orchards do, so all they gain from picking apples is familiarity and perhaps the support network of working with similar people.
The reason I mentioned the long run and a populist revolutionist because it takes many years for the labor force to adjust. An entire generation that can't find productive jobs for a decade is a very dangerous thing. This led to considerable political turmoil in the US in the 1930s. Unfortunately, I see something similar on the horizon in the service sector.
FFRobotics and Abundant Robotics have been demoing experimental systems for a while now. Both can pick apples, but both systems are still too slow and not field-ready yet. There have been working academic demo systems for about five years [1], but they're slow. Or slow and really complicated.[2]
Robot manipulation in unstructured situations still sucks. It can sort of be done, but doing it fast enough to be useful remains tough.
Just curious, what do you think the main bottlenecks are that cause physical slowness? The naive me asks, "Why can't you just turn up the speed/power/something?" but I assume it's more complicated than that or someone would have done it.
Sometimes the vision system is the bottleneck, but that's less of a problem today with abundant compute power. Sometimes it's that fast motion creates vibration in the system. Most of these systems are basically "find target, move to target". They're not continuously tracking the target and end effector as they get closer. Sometimes it's just weak mechanical design. Looking closely at those videos, you see duct tape and cable ties. They haven't done the mechanical engineering yet for a fast system.
For production, they need faster, cheaper, long life, maintainable, more adaptable to different plants, and pressure-washable. That's a lot of mechanical engineering.
As you can see this is some iterations away from being optimal but it is certainly getting there.
Primates eyes co-evolved with fruit evolving the colour-coded 'I am ready to eat' system they have to let us know when things are ripe. Imagine if the fruit picking machine gets to be reasonably aware of what fruits are ripe in the orchard at a given moment in time and is able to optimally pick fruit for a given customer, e.g. a sandwich bar chain may want ready to eat fruit whereas the supermarket may want something that ripens at home. A reasonable AI system could be trained to do a better job of reading the 'ripe' signals of fruit and even 'know' that given fruit since it was just a bud on a flower. Waste could be cut to a minimum if the system was aware of apples ready to go bad or drop to the floor.
As well as picking there is also the matter of grading the crop, getting it clean and so on. There is no reason why this machine can't do all of that at source, so the apple gets washed, blown spotlessly dry with a Dyson style dryer and then instantly sorted/graded and packed into some 100% nitrogen packet of some sort, sent to the distribution centre and on the shelves for lunch the next day.
Hopefully our new robotic overlords in the fields will work from above. Instead of heavy tractors, plant and people on the ground you could have some lighter weight equipment strung out from above, a mesh of wires held up by some massive wind turbine towers, to which the fruit picking machines might traverse.
McCain Chips are made with robots on the production line that pick out any defects, again, our fruit picking machines could do this and grade any cleaned produce as catering grade, nonetheless prepared clean and packed in nitrogen to prevent any further degradation. A whole new world of better product and time to market might be possible by handing over the fruit picking bit to the robots. Apples could even be picked when they got to a 'perfect size/weight' and not just graded that way, trees could be picked in such a way that all the fruit attains this 'ideal' which is optimal given inputs.
Such a system would work 24/7 in all seasons, tending the crop one way or another on a large estate, with some of its robotic arms going places at all times. You would need a team of maintenance and other support roles to keep things operational, all requiring a different skill set to the pickers of today. Their wages would be higher and spent in the local community rather than sent out of the country.
Orchards need not be so regimented for machines in this brave new world, orchards could have a mix of trees of all shapes and sizes, with different land use below, open to people and a proper 'forest floor' for small furry animals. Fruit in such areas could be expected to be part eaten by bugs but our robot overlords could also be experts at picking the bugs out to produce perfectly good fruit for pies and juices.
Rejecting our soon-to-be robot fruit picking machines is like rejecting the horse drawn plough as it might put the humans that pull ploughs out of business.
There is no optimal picking. Apples you buy in the store can be up to a year old. Washington apples are picked at harvest time and stored in refrigerated warehouses to be distributed nationwide throughout the year.
>> Members of the $7.5 billion annual Washington agriculture industry have long grappled with labor shortages Human pickers are getting scarce [..] “Young people do not want to work in farms, and elderly pickers are slowly retiring.”
>> The work is hard and dangerous, and has long drawn Mexican workers to central Washington [..] He estimated half the state’s farmworkers are immigrants who are in the country illegally [..] many of them have settled in Washington and are productive members of the community
>> ..the industry is deeply interested in alternatives to human labor..
>> The robotic pickers don’t get tired and can work 24 hours a day.
>> “They are scared of losing their jobs to mechanization,” [..] “A robot is not going to rent a house, buy clothing for their kids, buy food in a grocery and reinvest that money in the local economy.”