It we want to feed the world, when that world is throwing more extreme weather at us, we need to work out how to do companion planting at scale. (think how east coast indians did farming) IF we can make practical farm robots, we can not only remove the need for herbiscides (direct manual intervention, ie physically weeding buy pulling out the seedlings) but also keep ground cover even after cropping, meaning much less water loss.
Soil degradation is a real threat. the way we farm now means we have massive monocultures, large tracks of land that are bare for weeks on end. All of this requires lots of inputs to be productive. The promise of non-pesticide farming is that you get much richer soil, because you're not killing off the stuff that lives there.
But we need to understand what makes a soil productive, however that changes based on location and crops.
But how do you kill the cover crop so you can grow wheat again? How do you kill the weeds? The answer for hundreds of years has been ploughing, but that is exactly what we are trying to avoid. The only viable answer today is...Roundup (glyphosate).
And there is the rub. To farm with better soil health, and less ploughing today requires a chemical that we are not happy with using.
A robot to pull the wild oats out of a wheat field sounds practical. A robot to pull 100 acres of white mustard and weeds is what?
There is some work with special rollers that can kill leafy cover crops, and there are tractor pulled mowers, of course, but it is a partial solution. Afterwards you still have a field of dandelions and black grass. So they use roundup.
Then there is the break crop issue. After wheat you would plant rape or beans, perhaps, but only rape will make you a profit, but this is a tremendous risk. A flea beetle outbreak will kill the entire crop. The solution until recently was neonicitinoid coated seed, but that is now banned. So what do you grow?
Part of the solution for me is mixed farming. Wheat followed by fodder beet, graze it off by sheep. Also the drought tolerant lucerne (Alfalfa to the rest of the world). Then seed grass and put cows or sheep (and hat and silage) followed by poultry (bird flu dependent) in a paddock grazing system. Then plough it and back to wheat.
Smaller automated machines could allow smaller fields and a more diverse patchwork I suppose. Cooperation needs to increase massively between farms, so a dairy farm partners with a arable farmer on one side and a sheep farmer on the other.
All depending on your soil type and topography of course. Lots of ground is grazing only.
Then you need to make the economics work. Small farms don't pay. Of course 1000 acres of mountain ground is totally different to 1000 acres of flat arable ground.
We definitely need some innovative in the economics. The current model of subsidy is laughable. Farmers being incentivised to grow no crops at all can't be the answer to food security!
I would love to work in this sector. I feel with better automation and better economics we could make smaller farms that are more like a market garden with many different crops could work. A practical (and cheap) way to harvest grain on a small field would be the biggest breakthrough for me.
I don't agree, and I note that you also answered your question differently later in your post with the note about 'mixed farming' (grazing it off).
There are, of course, other answers than herbicides. Seasonal crops, harvesting and then seed-sowing amongst the stubble (provides some mulch & eroson protection), intensive strip-grazing (bovine, ovine, caprine, or fowl, all effective options), or even a cycle or two of fallow.
Broadscale monoculture invites its own range of problems, and herbicide-resistant highly-competitive (when in a single-other-species ecosystem) weeds are one such.
If the starting position is 'we must grow the same variety of wheat in the same field every year', then indeed, you're going to have some challenges.
I don't really think its possible yet, (or it might not be possible with the current breeds of wheat)
Lets step back and say: could a bunch of humans with hand tools do this?
Yes, they could. (As opposed to, say, eradicating horsetail: if we can't even do it ourselves, then no, we definitely aren't getting a robot to do it).
One way could be the lasers-on-trailer approach, burn the undesirable plants so they're sufficiently uncompetitive. Another could have arms and cutters to reach down and sever the plant below the surface.
Either of the above could instead be a smaller autonomous robot working as part of a swarm day and night.
Then the grass will go to seed and spread.
Then if the grass seeds got into next year's wheat seed you will have a very big problem in future!
Roller crimpers used at the right time can sometimes be effective at killing a cover crop and using it as a mulch to prevent weeds.
Others have mentioned roller crimping. There's hoeing, flame weeding, and others. Plenty of rabbit holes for you
Edit: addressing your second point, flame weeders and hoes are not really effective on perennial grasses. With scutch (couch grass) hoes just divide it up. Burning the leaves off doesn't kill it either, it just grows again. They used to plough and then harrow the roots out repeatedly, burning yet more energy.
My first offered technique was occultation not crimpers. That of course won't work in all circumstances. Often you have to use multiple techniques and really stay on top of the persistent ones. Automation can do that but then there is the pricing issue.
As noted by others you seem quite knowledgeable for someone not in the field but also wrote confident that you have the answers.
[0] https://hortsense.cahnrs.wsu.edu/fact-sheet/weeds-quackgrass...
I have found it rather disappointing in practice, in my experiments. It needs to stay down for a very long time to kill bindweed, for instance. You would think a month would do it, but it won't. :(
https://naturalhistory.si.edu/research/botany/news/plant-pre...
https://www.earth.com/news/native-fungi-native-trees-plants-...
https://matjournals.net/pharmacy/index.php/IJPPR/article/vie...
Science is not reality. We abstract reality to make nice and useful models. Reality violates our models constantly.
If you want to go to the furthest defensible extreme[0], then all of life is just one long, violent, ever-expanding chemical chain reaction, that started some 4+ billion years ago, and shows no sign of stopping[1]. What we call life - cells and plants and fungi, bacteria and cows and people - are just stable-ish substructures you could identify within the fractal complexity of that chemical fire, that completely enveloped the surface of this planet, cracked it and reached deep underneath, and recently even started spitting bits of itself to the Moon, Mars, and even beyond the Solar System.
This framing isn't particularly useful to us most of the time, but I find that occasionally invoking it helps really understand that there is no such thing as "an individual" or "a specific object" in the physical universe, no true boundary separating this fox from that squirrel, or this person from another. It's how we perceive the world because the approximation holds up at our time scales, but on occasion (such as when discussing nano-scale things, or evolutionary biology), it's worth remembering it's not true. Nature doesn't have boundaries.
--
[0] - Going further makes things too generalized to be useful. Like, yes, we're all made from star stuff.
[1] - Think of it like of the "Game of Life" or such simulations, where most states quickly decay to nothing or some static form, but every now and then you'll find a configuration that just explodes and keeps going, expanding its borders and perhaps leaving behind some further explosions on a fuse, recursively. Life is like that.
You find it difficult to accept, or is it just your brain that finds it difficult to accept?
I was afraid that would happen. My comment was really more aimed at being a comment to yours, than a reply. The fact that you're starting to "feel" this as being more true is not negated or impacted by it being an existing thought. Thoughts like this take time to settle into experienced truth, and i appreciate that. Had we been conversing that would have been a non-sequitur, and i would not have made it.
One of the problems with comment systems though is that we are at once conversing and broadcasting. The comment was more intended on being a broadcast than a direct reply to you, as a breadcrumb for anyone interested in the path you were taking to maybe seek it out in existing literature.
Once one mostly outgrows it, things are much easier and less tense. Even if a discussion dynamic increases the probability for defensiveness.
Ego is the enemy.
But I’m more of a farmer at heart. For my newsletter/platform I follow agricultural research from the Netherlands on a daily basis. That country is arguably the best center for agricultural R&D anywhere. This particular study is evidence of that.
That makes HN threads on farming equally painful to read. The “why don’t they just,” “we should do more about,” and “we need to innovate/research” takes are almost always mis- or superficially informed.
The term “mycorrhiza” for example was coined by Albert Bernhard Frank in 1885. So we’ve been doing research on this topic for at least 14 decades already.
Definitely far ahead on the curve on "how to grow things efficiently in a greenhouse", but then there are usually giant heaters next to those..