RoboCrop: Teaching Robots How To Pick Tomatoes (phys.org) 29
alternative_right quotes a report from Phys.org: To teach robots how to become tomato pickers, Osaka Metropolitan University Assistant Professor Takuya Fujinaga, Graduate School of Engineering, programmed them to evaluate the ease of harvesting for each tomato before attempting to pick it. Fujinaga's new model uses image recognition paired with statistical analysis to evaluate the optimal approach direction for each fruit. The system involves image processing/vision of the fruit, its stems, and whether it is concealed behind another part of the plant. These factors inform robot control decisions and help it choose the best approach.
The model represents a shift in focus from the traditional 'detection/recognition' model to what Fujinaga calls a 'harvest-ease estimation.' "This moves beyond simply asking 'can a robot pick a tomato?' to thinking about 'how likely is a successful pick?', which is more meaningful for real-world farming," he explained. When tested, Fujinaga's new model demonstrated an 81% success rate, far above predictions. Notably, about a quarter of the successes were tomatoes that were successfully harvested from the right or left side that had previously failed to be harvested by a front approach. This suggested that the robot changed its approach direction when it initially struggled to pick the fruit. "This is expected to usher in a new form of agriculture where robots and humans collaborate," said Fujinaga. "Robots will automatically harvest tomatoes that are easy to pick, while humans will handle the more challenging fruits."
The findings are published in Smart Agricultural Technology.
The model represents a shift in focus from the traditional 'detection/recognition' model to what Fujinaga calls a 'harvest-ease estimation.' "This moves beyond simply asking 'can a robot pick a tomato?' to thinking about 'how likely is a successful pick?', which is more meaningful for real-world farming," he explained. When tested, Fujinaga's new model demonstrated an 81% success rate, far above predictions. Notably, about a quarter of the successes were tomatoes that were successfully harvested from the right or left side that had previously failed to be harvested by a front approach. This suggested that the robot changed its approach direction when it initially struggled to pick the fruit. "This is expected to usher in a new form of agriculture where robots and humans collaborate," said Fujinaga. "Robots will automatically harvest tomatoes that are easy to pick, while humans will handle the more challenging fruits."
The findings are published in Smart Agricultural Technology.
Or alternatively... (Score:1)
... use the 1.85 million unemployed in Japan to pick the tomatoes. I doubt they're all wannabe rocket scientists or AI devs just waiting for their break.
Re: (Score:2)
Every economy will always have some unemployed and underemployed people, as every economy is constantly shifting, as does people's status as to what kind of work they can carry out. That doesn't mean it's efficient or sustainable to say, "Oh, you lost your job yesterday? You're a farmer now!". Japan has an unemployment rate of ~2,5%, which is actually quite low. Japan is famous for making "busywork" jobs for people. "Oh, you have a traumatic brain injury and have trouble walking quickly? Just stand over
Re: (Score:2)
... use the 1.85 million unemployed in Japan to pick the tomatoes. I doubt they're all wannabe rocket scientists or AI devs just waiting for their break.
The problem in Japan is that they restrict immigrant workers even more than the US. Just like most Americans, most Japanese wouldn't want to pick tomatoes (or any other farm product) because the wages are far too low for the difficulty of the work. Even unemployed Americans and Japanese would refuse to pick tomatoes for such a low wage accompanied by literally back-breaking work. That's why Trump's immigration crackdown is affecting American fruit and vegetable farmers so severely.
The only ways to combat
How many could? disabled/elderly? + racist culture (Score:2)
... use the 1.85 million unemployed in Japan to pick the tomatoes. I doubt they're all wannabe rocket scientists or AI devs just waiting for their break.
Good point, but toiling in the fields is back-breaking work. What percentage of Japan's unemployed could physically work half a shift on a farm? In the USA, a huge portion of the unemployed are disabled. I don't think having 60 year olds or people with multiple sclerosis working the fields is a great idea. Also, you have to think long term...are the unemployed in Japan increasing or decreasing in age? Is their median age increasing or decreasing?...or more directly, even if you could find enough from t
Good thing (Score:2)
So robots are the new fat cat capitalist class (Score:3)
Re: (Score:2)
Leave it to Slashdotters to complain about tomato-picking going from "repetitive mindless activity at maximum velocity" to "only the cases that you have to use your brain to think about because a robot isn't smart enough".
I guess we should have people manually screwing the caps onto toothpaste tubes also.
People are cheap (Score:4, Insightful)
I worked for a farming automation company over 20 years ago now. There were a few things I noticed:
The first was that much of the low hanging automation tasks had already been automated a long time ago. People think automation is replacing a field full of 100 workers with 100 humanoid robots. But the reality is that we replaced those workers with a tractor and pesticide sprays. It's this observation that makes me skeptical about the whole humanoid hype fest.
The second thing is that people are damn cheap. I mean, a human can pick a lot of tomatoes in an hour. If the human breaks down, you just fire them and hire one that is in better condition. There is no capital cost for a human (perhaps a little to train them) - the farm doesn't have to pay to 'build' them. Even if a tomato picking robot was a few 1000's of dollars (not going to happen) that would still be higher than the cost of just getting another human. Further, if markets change you just fire your humans, or get them to pick something else instead, but if you've invested significant capital in tomato picking robots you've got a big problem.
I'm not saying that there isn't a point at which an automatic tomato picking robot wouldn't be viable - there definitely will be. But ultimately it's just an economics question. At the moment, making such a robot that can even perform that task well, let alone be cheap and, importantly, reliable, is a very difficult problem. I definitely think we could solve it - we could have solved it a decade ago - but there is very little investment for this stuff because the low price of humans sets a limit on the value of the resultant product, and that value is very low.
Re:People are cheap (Score:5, Informative)
You have a strange definition of humanoid. [b-cdn.net]
Depends entirely where you are. Labour in our greenhouses here (Iceland) is insanely expensive and makes it difficult to compete on the global market against imports (even imports of perishable things that are expensive to ship). Replacing workers with machines on a given task might not be economical, in, say, Uganda, while it might be a complete no-brainer here.
That said, there are also things that we could do to bring labour costs down, but don't. For example, at the same time we struggle with high labour costs and shortages in fields like agriculture, we also have problems with too many asylum seekers, most of whose cases get rejected and kicked out (economic migrants), but not for many months of limbo, living at taxpayer expense, when what they really want to do is work here. The obvious solution is to make an agricultural worker visa, where while you're guaranteed the same labour/safety standards but you're not guaranteed Icelandic minimum wages and benefits - but still far better than these people would get in their homeland. The vast majority of them would be thrilled to sign up for such a thing if it would guarantee them residence. But migration-politics is such a hotbutton political issue right now globally, it's hard to do any commonsense stuff like that.
I'd also add that while picking is the glamorous AI task, it's only part of the work. In greenhouses at least (I don't have field-tomato cultivation experience) we do an awful lot to manage the plants**. You have to remove the suckers at every internode on the plant (except those at the base to get the proper number of vines per plant), you have to remove the lower leaves at regular intervals, you have to wind the plants around the string that they grow up, when the plants get tall you have to slide the tops over so that they grow at a diagonal, and on and on. And that's just the management for the mature plants - you also replace your plants at regular intervals, so you have to start new plants grow them, repot them, grow them, then plant them out (not just the planting, but also replacing the growing medium) - plus all of the side stuff like cleaning, managing irrigation, and on and on. Harvesting the fruit is just one task among many.
** To anyone reading this who is surprised about all of the plant manipulation, think of it this way: you have a finite amount of surface area for light to hit, and a finite amount of root volume for each plant. So once your plants get to a size where they're basically using all the light and basically have rooted through the whole growing medium, the only way you can keep their growth in balance as they continue to grow is to keep removing old leaves. And you need to stop any branching immediately because again all that branches would do is just grow into your other plants. And once your plants are so long that they've hit the top of your wires, the only thing you can do is slide the wires over so that the plants are growing (ever-increasingly) diagonally. Nothing you can do about the fact that the stems just keep getting longer and longer as the top continues to grow except replace the plants at regular intervals - you need to let the plant continue to grow because that's how you get new blossoms for fruiting. Removing the lower leaves also has the nice side effect that the tomatoes that are maturing lower down are left fully exposed [novagric.com], making them easy to harvest. You harvest as th
Yeah but they keep trying to form unions (Score:1)
So the real problem with Farm labor is going to be that nobody is going to want to pay people enough to have people who want to live on a farm if you're not picking up refugees from war-torn countries. And a lot of people don't want those refugees anymore for a variety of reasons.
Otherwise you've got to pay peop
Re: (Score:1)
LaserWeeder + Robot strengths != human ones (Score:3)
Re: (Score:2)
[William Stanley Jevons has entered the chat]
Re: (Score:2)
You say that as if there is an upward mobility path, start by working on the farm, then...where do people "graduate to"? Do they gain skills that would apply to higher skill jobs? That's the danger for society overall in AI, because people do need to start SOMEWHERE to learn the more advanced skills that AI can't do. Yea, you need a doctorate just to get a job, but those with a high level degree still won't have the job experience needed to do well in many jobs where experience teaches better than any
Re: Keep jobs for immigrants (Score:2)
Would be interesting to know how many legal immigrants these jobs.
As for the illegal ones....they are allowed to be here. Not letting them take these jobs isn't cruelty, it's enforcing the law.
Re: (Score:2)
Would be interesting to know how many legal immigrants these jobs.
Looks like the estimates are at about 40% https://www.ers.usda.gov/data-... [usda.gov]
Re: (Score:2)
Oops, that's 40% illegal immigrants. It's 60% people in the country legally.
Re: Keep jobs for immigrants (Score:2)
So, it would APPEAR that there are a large number of these jobs that would've available to legal immigrants if it were notnfornillegalnones taking those jobs.
an 81% success rate...still not great (Score:2)
RoboCrop, coming soon to Old Detroit (Score:2)
1. Serve the public taste
2. Protect the delicate flesh
3. Uphold the hoe
This is the future of agriculture (Score:2)
Imagine when each tomato is picked only at the time of optimum ripeness. The robot will be back every day to check the plant.
Imagine wine made from grapes that are individually picked at optimum sweetness and acidity.
Imageine pest control being mechanically done with robots individually crunching each pest; no pesticides.
Cotton being picked such that every single strand is harvested.
Orchard fruit being automatically inspected and sorted for composting, juicing, or retailing. Perhaps with radar or sonar.
Seed
How many tomatoes? (Score:2)
Just how many tomatoes can a person pick, and how much are they paid for that?
I bought tomatoes yesterday. 16 tomatoes weighing 250g and costing £1. Do your own conversions.
How much of the retail price is the picking cost? Minimum wage is about £12 or something and I could pick a heck of a lot of tomatoes in an hour.