4 ms·

An additional issue is that a good fraction of 'organic nitrogen fertilizer' comes in the form of fish meal. This isn't a very environmentally friendly or susta

An additional issue is that a good fraction of 'organic nitrogen fertilizer' comes in the form of fish meal. This isn't a very environmentally friendly or sustainable approach, however - e.g. at peak extraction, 2/3 of sardine fishery production on the US west coast was being converted into crop fertilizer (mid 20th century). Note however that modern industrial agriculture uses fertilizer fairly inefficiently, and a large fraction just runs off the fields into waterways. Here is where the AI robots could make a big difference, by crawling up and down farm fields and selectively applying fertilizer to individual plants as needed (while also removing weeds and treating pest infestations). Of course, this doesn't make agrichemical producers all that happy as it could greatly reduce demand.

searine 3y ago
>Here is where the AI robots could make a big difference Usually when someone says "AI will save the world because" it is followed by a bunch of naive hyperbole. However, you are 100% right. A few robust robots with an AI to spot pests and poor nutrition could decimate input costs for a farm and solve several environmental issues in one go.
- snapcaster 3y ago
  What are you basing that on?
  - searine 3y ago
    The reason chemical pest control and broadly spraying fertilizer on the top of fields are a thing is that they have low labor costs. You'd get more efficacy if you walked the field and removed pests by hand or sprayed fertilizer only where you need it. It's the simple reason why a well-tended garden produces better produce than a corporate farm. There is simply more effort per plant in a garden. Tractors are currently too dumb to treat plants individually. They require plants to emerge, grow, harvest at the same time, and in the same way. Current modalities for pests/nutrition treat plants at the field level, not at the individual level. This creates lots of waste and externalities. If there were a robustly built robot smart enough to deal with plant issues in and individual way then there is a huge slice of efficiency that could be gained. A field with the productivity of a garden. People have been trying to create this for a decade now and the limiting factor is the software. Maybe with the latest round of AI advances that will change.
    - matthewdgreen 3y ago
      For example: https://carbonrobotics.com/ https://carbonrobotics.com/
cmrdporcupine 3y ago
You don't need robots -- tractors already drive up and down the field, and the cost of the tractor is many times the annual cost of the worker, BTW -- you just need good application. Which is something farmers already attempt to do. Because why would you waste your $$ on bad deployment of expensive fertilizer? (Or pesticide, etc.) This is my employer's bread and butter for many decades: https://www.ravenind.com/products/applications-booms https://www.ravenind.com/products/applications-booms And yes, it's a lot of emphasis on autonomy now, but you'd be amazed at what just having better spray nozzles can do? Cash crop farming is an incredibly capital intensive operation. Robots don't fix that.
- steve_adams_86 3y ago
  Your point about spray nozzles is a great one. Not many people realize how difficult it is to make such a small and seemingly unimportant component work well and reliably. In my hydroponics system, spray nozzles are one of the more expensive (for the size and purpose) components and they fail the most often. You’d think a brass nozzle would last years, but it’s going to need to be disassembled and cleaned before a year is up. Now imagine a huge tractor with high pressure bars that are lined with dozens of nozzles operating in a dirty environment and moving huge volumes of fluid. You’re going to get clogs, you’re going to have to clean them, and the application will be non-ideal quite often if you don’t. And that’s just the nozzles. This stuff is so much harder and more complicated than it seems on the surface. Adding automation almost makes it harder because you really do want a human observing the system and ensuring it’s actually working.
  - foobarian 3y ago
    Neither here nor there, but why are specialized nozzles needed for fertilizer? Can't that go directly on the ground with a very basic pattern?
    - cmrdporcupine 3y ago
      There all sorts of things that affect coverage. These things are done by booms with nozzles at all sorts of different points, and there's variables like pressure variance at different points in the pipes, spray radius per nozzle type, turning radius / position of the tractor, relative height of the boom to the crop / ground at different points, etc. And it's not just fertilizer, but other sprays, too. Here, this looks thoroughly boring, but this guy is going over configuring one of our systems for fertilizer application... You can get an idea of some of the variables involved, if you can make it through :-) https://www.youtube.com/watch?v=2rQJI7zz1q4 https://www.youtube.com/watch?v=2rQJI7zz1q4

searine 3y ago

>Here is where the AI robots could make a big difference Usually when someone says "AI will save the world because" it is followed by a bunch of naive hyperbole. However, you are 100% right. A few robust robots with an AI to spot pests and poor nutrition could decimate input costs for a farm and solve several environmental issues in one go.

snapcaster 3y ago
What are you basing that on?
- searine 3y ago
  The reason chemical pest control and broadly spraying fertilizer on the top of fields are a thing is that they have low labor costs. You'd get more efficacy if you walked the field and removed pests by hand or sprayed fertilizer only where you need it. It's the simple reason why a well-tended garden produces better produce than a corporate farm. There is simply more effort per plant in a garden. Tractors are currently too dumb to treat plants individually. They require plants to emerge, grow, harvest at the same time, and in the same way. Current modalities for pests/nutrition treat plants at the field level, not at the individual level. This creates lots of waste and externalities. If there were a robustly built robot smart enough to deal with plant issues in and individual way then there is a huge slice of efficiency that could be gained. A field with the productivity of a garden. People have been trying to create this for a decade now and the limiting factor is the software. Maybe with the latest round of AI advances that will change.
  - matthewdgreen 3y ago
    For example: https://carbonrobotics.com/ https://carbonrobotics.com/

cmrdporcupine 3y ago

You don't need robots -- tractors already drive up and down the field, and the cost of the tractor is many times the annual cost of the worker, BTW -- you just need good application. Which is something farmers already attempt to do. Because why would you waste your $$ on bad deployment of expensive fertilizer? (Or pesticide, etc.) This is my employer's bread and butter for many decades: https://www.ravenind.com/products/applications-booms https://www.ravenind.com/products/applications-booms And yes, it's a lot of emphasis on autonomy now, but you'd be amazed at what just having better spray nozzles can do? Cash crop farming is an incredibly capital intensive operation. Robots don't fix that.

steve_adams_86 3y ago
Your point about spray nozzles is a great one. Not many people realize how difficult it is to make such a small and seemingly unimportant component work well and reliably. In my hydroponics system, spray nozzles are one of the more expensive (for the size and purpose) components and they fail the most often. You’d think a brass nozzle would last years, but it’s going to need to be disassembled and cleaned before a year is up. Now imagine a huge tractor with high pressure bars that are lined with dozens of nozzles operating in a dirty environment and moving huge volumes of fluid. You’re going to get clogs, you’re going to have to clean them, and the application will be non-ideal quite often if you don’t. And that’s just the nozzles. This stuff is so much harder and more complicated than it seems on the surface. Adding automation almost makes it harder because you really do want a human observing the system and ensuring it’s actually working.
- foobarian 3y ago
  Neither here nor there, but why are specialized nozzles needed for fertilizer? Can't that go directly on the ground with a very basic pattern?
  - cmrdporcupine 3y ago
    There all sorts of things that affect coverage. These things are done by booms with nozzles at all sorts of different points, and there's variables like pressure variance at different points in the pipes, spray radius per nozzle type, turning radius / position of the tractor, relative height of the boom to the crop / ground at different points, etc. And it's not just fertilizer, but other sprays, too. Here, this looks thoroughly boring, but this guy is going over configuring one of our systems for fertilizer application... You can get an idea of some of the variables involved, if you can make it through :-) https://www.youtube.com/watch?v=2rQJI7zz1q4 https://www.youtube.com/watch?v=2rQJI7zz1q4