It is a hard problem. Historically, people often grew plants in heterogeneous environments because they grew their own crops. Even if it wasn't _all_ of their food, it was still a subset of their food. But it was very labor intensive.

The scale of monocrop farming today with modern tractors is a very new field as well (no pun intended).

With machine learning specifically, I can see a robot arm being able to farm a lot of mixed crops. This is already happening for the most labor-intensive crops like strawberries[0]. Robots are cheap and can work 24/7.

I think it just a matter of time before computers advance enough to be able to perform the basic, repetitive tasks that humans are able to.

0: https://www.therobotreport.com/strawberry-picking-robot-make...

This very interesting piece might be of interest for you : https://99percentinvisible.org/episode/how-to-pick-a-pepper/

It took one of the protagonist ten years of testing to find a machine capable of harvesting special-bread peppers. The breakthrough your are looking for isn't for software engineering only - or even at all -. It's a combination of mechanical engineering, biology and hacking. Venture-led startups often lead to short-lived niche products made to impress silicon valley. It's not sustainable, and it's not the future.

> don't you think humans already tried that in the past 10 000 years ?

Often what you are saying is true. Except in this case we have access to new technologies that have only existed for a small percentage of that 10,000 years. It might be worth re-examining what we think we know about growing food in light of these new technologies. That is, things that would have been ideal but were practically impossible may now be possible.

Software gives the agriculture industry new tools, it helps gather information, it helps build better machinery and can even help train people. Of course software engineers won't be the only ones contributing to solving these problems, but farmers are very resourceful and tend to welcome any new tools to improve their businesses. They've been improving for the last 10,000 years, why stop now?