4 ms·

Low wages for field workers (e.g. picking berries) has stunted automation of these tasks, though I think we're now at the level of robotics (especially computer

Low wages for field workers (e.g. picking berries) has stunted automation of these tasks, though I think we're now at the level of robotics (especially computer vision) where it's feasible.

DethNinja 7y ago
We are definitely at this stage. So long as ruling class wishes they can make huge automation improvements in agriculture in just couple of decades and easily feed the current world population.
Etheryte 7y ago
Sadly, I think this is an example of what I would maybe call the HN Dunning-Kruger approach to many problems: "just" use IT/AI/robots/etc. There are many companies who have been working in this space for a long time, there's a reason why it's still manual work — it's freaking hard. Berries are a good example of this, personally I'm intricately familiar with the strawberry industry, but I believe it applies elsewhere too. Strawberries are very tender. If you apply too much pressure, it will mulch and one berry going bad is usually enough to get the whole box moldy. The second tricky bit is that you want to pick strawberries with the right amount of stem remaining: too little and you risk hurting the berry, too much and the stem will break other berries in the box, leading to mold again. One final difficult problem is finding the berries: strawberries often grow fairly bushy, especially so if you grow one of the varieties that gives large berries. To find the strawberries in the bush, you usually gently move around the leaves to see if there are any. This is again a very "feel" based movement, too little movement and you miss the berry, too much and you damage the plant. I'm not saying this won't ever be done. As mentioned before, there are companies who are actively working in this space, many of them for many years already. I do think however that the problem is considerably more difficult than what may be observed at first.
- jiggawatts 7y ago
  It has been done! https://www.youtube.com/watch?v=FCivsotZEjk https://www.youtube.com/watch?v=FCivsotZEjk There's actually several designs out there for picking strawberries, but like the OP said, expensive robots aren't competitive with cheap labour. If the labour pool dries up, they're more economically viable.
  - Etheryte 7y ago
    There's a very big difference between a proof of concept "can pick some berries" and a solution that's viable at scale, even if ignoring financial limitations. The video edgewise admits to it as well: > The country's largest supplier of berries has made use of the robot at one of its berry fields. The operation shown in the video is likely not viable for actual pick and sell with a number of the issues I outlined above. As I said, there are many companies working on it, but claiming it's a solved problem is simply handwaving away the truth.
  - s1artibartfast 7y ago
    >If the labor pool dries up, they're more economically viable. I wonder if strawberries would cease being an economically viable crop before this level of mechanization becomes viable.
- jtolmar 7y ago
  Like all robotics, a solution needs someone willing to develop expert domain knowledge, not a generalist software engineer with an enthusiasm for the latest out-of-the box AI techniques. And part of that expert domain knowledge is almost certainly going to be knowing to start with something easier than strawberries. My best guess is apples, but I'm not that expert. But, since you provided a thoughtful analysis of why strawberries are hard, here are some thoughts on the problems you mentioned (to be clear, I don't think I'm solving strawberry picking in a random HN comment, I just thought it was a thought-provoking problem to discuss): > If you apply too much pressure, it will mulch My robotics knowledge is super out of date, but AFAIK applying a reasonable amount of pressure is stupid hard. Can you avoid ever touching the strawberry and just always grab it by the cut stem? > The second tricky bit is that you want to pick strawberries with the right amount of stem remaining Always grab too much stem, then trim it down on the way to the bin, so you can do this on a known background instead of the middle of a green mess. The trimming cutter may even be designed to round down the stem. > This is again a very "feel" based movement, too little movement and you miss the berry, too much and you damage the plant. Can you get away with combing the entire plant to the side, or is the plant fragile too? Failing that, a robot can stick its eyes right in there and scan every strawberry-sized voxel of space.
ilikepi 7y ago
There was a recent Planet Money episode[1] about this. [1]: https://www.npr.org/2020/03/25/821593542/episode-984-food-and-farmworkers https://www.npr.org/2020/03/25/821593542/episode-984-food-an...
jeffchien 7y ago
It's not simple as that, and that's a problem that people have been working on for a long time. Some crops need to be reengineered to be friendly to mechanical pickers: https://99percentinvisible.org/episode/how-to-pick-a-pepper/ https://99percentinvisible.org/episode/how-to-pick-a-pepper/ And even then the yield is bad: * https://www.npr.org/sections/thesalt/2018/03/20/592857197/robots-are-trying-to-pick-strawberries-so-far-theyre-not-very-good-at-it https://www.npr.org/sections/thesalt/2018/03/20/592857197/ro... * https://www.growingproduce.com/fruits/berries/florida-blueberry-growers-fielding-rise-of-the-machines/ https://www.growingproduce.com/fruits/berries/florida-bluebe... * http://www.beilkefamilyfarm.com/blog/2015/7/blueberries-machine-picking http://www.beilkefamilyfarm.com/blog/2015/7/blueberries-mach...

Low wages for field workers (e.g. picking berries) has stunted automation of these tasks, though I think we're now at the level of robotics (especially computer vision) where it's feasible.

DethNinja 7y ago

We are definitely at this stage. So long as ruling class wishes they can make huge automation improvements in agriculture in just couple of decades and easily feed the current world population.

Etheryte 7y ago

Sadly, I think this is an example of what I would maybe call the HN Dunning-Kruger approach to many problems: "just" use IT/AI/robots/etc. There are many companies who have been working in this space for a long time, there's a reason why it's still manual work — it's freaking hard. Berries are a good example of this, personally I'm intricately familiar with the strawberry industry, but I believe it applies elsewhere too. Strawberries are very tender. If you apply too much pressure, it will mulch and one berry going bad is usually enough to get the whole box moldy. The second tricky bit is that you want to pick strawberries with the right amount of stem remaining: too little and you risk hurting the berry, too much and the stem will break other berries in the box, leading to mold again. One final difficult problem is finding the berries: strawberries often grow fairly bushy, especially so if you grow one of the varieties that gives large berries. To find the strawberries in the bush, you usually gently move around the leaves to see if there are any. This is again a very "feel" based movement, too little movement and you miss the berry, too much and you damage the plant. I'm not saying this won't ever be done. As mentioned before, there are companies who are actively working in this space, many of them for many years already. I do think however that the problem is considerably more difficult than what may be observed at first.

jiggawatts 7y ago
It has been done! https://www.youtube.com/watch?v=FCivsotZEjk https://www.youtube.com/watch?v=FCivsotZEjk There's actually several designs out there for picking strawberries, but like the OP said, expensive robots aren't competitive with cheap labour. If the labour pool dries up, they're more economically viable.
- Etheryte 7y ago
  There's a very big difference between a proof of concept "can pick some berries" and a solution that's viable at scale, even if ignoring financial limitations. The video edgewise admits to it as well: > The country's largest supplier of berries has made use of the robot at one of its berry fields. The operation shown in the video is likely not viable for actual pick and sell with a number of the issues I outlined above. As I said, there are many companies working on it, but claiming it's a solved problem is simply handwaving away the truth.
- s1artibartfast 7y ago
  >If the labor pool dries up, they're more economically viable. I wonder if strawberries would cease being an economically viable crop before this level of mechanization becomes viable.
jtolmar 7y ago
Like all robotics, a solution needs someone willing to develop expert domain knowledge, not a generalist software engineer with an enthusiasm for the latest out-of-the box AI techniques. And part of that expert domain knowledge is almost certainly going to be knowing to start with something easier than strawberries. My best guess is apples, but I'm not that expert. But, since you provided a thoughtful analysis of why strawberries are hard, here are some thoughts on the problems you mentioned (to be clear, I don't think I'm solving strawberry picking in a random HN comment, I just thought it was a thought-provoking problem to discuss): > If you apply too much pressure, it will mulch My robotics knowledge is super out of date, but AFAIK applying a reasonable amount of pressure is stupid hard. Can you avoid ever touching the strawberry and just always grab it by the cut stem? > The second tricky bit is that you want to pick strawberries with the right amount of stem remaining Always grab too much stem, then trim it down on the way to the bin, so you can do this on a known background instead of the middle of a green mess. The trimming cutter may even be designed to round down the stem. > This is again a very "feel" based movement, too little movement and you miss the berry, too much and you damage the plant. Can you get away with combing the entire plant to the side, or is the plant fragile too? Failing that, a robot can stick its eyes right in there and scan every strawberry-sized voxel of space.

ilikepi 7y ago

There was a recent Planet Money episode[1] about this. [1]: https://www.npr.org/2020/03/25/821593542/episode-984-food-and-farmworkers https://www.npr.org/2020/03/25/821593542/episode-984-food-an...

jeffchien 7y ago

It's not simple as that, and that's a problem that people have been working on for a long time. Some crops need to be reengineered to be friendly to mechanical pickers: https://99percentinvisible.org/episode/how-to-pick-a-pepper/ https://99percentinvisible.org/episode/how-to-pick-a-pepper/ And even then the yield is bad: * https://www.npr.org/sections/thesalt/2018/03/20/592857197/robots-are-trying-to-pick-strawberries-so-far-theyre-not-very-good-at-it https://www.npr.org/sections/thesalt/2018/03/20/592857197/ro... * https://www.growingproduce.com/fruits/berries/florida-blueberry-growers-fielding-rise-of-the-machines/ https://www.growingproduce.com/fruits/berries/florida-bluebe... * http://www.beilkefamilyfarm.com/blog/2015/7/blueberries-machine-picking http://www.beilkefamilyfarm.com/blog/2015/7/blueberries-mach...