Hacker News Reader

Comment by dhbradshaw | Hacker News Reader

I think one of the reasons to get bigger and bigger is to save human time.

If we get to a place where it's all AI, is there a chance that many smaller machines will become more effective than a few enormous ones?

View on HN

In fact no. Bigger heavy machines do less damage than smaller ones overall. You have to look at the whole field, not just where the tires touch. Where the tires touch the ground the heavy machine is worse, but the smaller machines touch the ground in a lot more places and so do more damage.

Farmers are now using GPS to ensure all the tires that touch the field drive exactly the same place every pass, every year. Where the tires touch the ground is hardly worth farming, but the rest of the ground is undisturbed and so much healthier.

> smaller machines touch the ground in a lot more places an so do more damage.

That’s certainly possible but not a given, and I think the somewhat obvious implicit suggestion was to use smaller machines with a lower weight to surface area ratio, no? Damage isn’t a function of how many places the ground is touched at all, it’s only a function of weight per unit area.

Is it possible we could design small robots that weigh less per unit area than the large tractors? Sure, why not? We could have smaller machines with bigger tires, we could design machines with weight reducing features like propellers, we can choose to use lighter weight materials & designs for machine frames & engines. It seems like there a plenty of possibilities that are not in fact impossible.

The function multiples weight by a very small factor though, so area dominates. (I have no idea what the function is, I've just had conversations with soil experts who tell me this)

What area? What weight? We need specifics in order to make claims one way or the other. It’s not just possible, but easy to design machines with lower weight to surface area ratio, so it’s just wrong to conclude the answer to the question can do we better is somehow no.

You can get a phd answering this question, something I do not have. What I have done is talk to such phd's - 5 years ago. I'm not giving more details because I don't remember more. I remember this much because it stood out to me as so non-obvious.

I would think so. There is a lot going on in this area, see e.g. https://pixelfarmingrobotics.com/robot-one/ and https://www.odd.bot/

Doesn't the economy of scale teach us that one big apparatus is cheaper than many smaller ones?

The soil doesn’t care how much it costs.

Stop thinking with your wallet; our times demand a more holistic approach.

Scale is not size.

"Economy of scale" is not a universal law as CEOs (and some economists) like to pretend.

That first one is incredible and the second looks like a more cost effective option. Most don’t realize that organic farming isn’t pesticide free but this gives farmers that option without resorting to hand-weeding.

There's also the trade-off between width of the worked area and the treaded surface made unusable for the season. If you have any tractor work steps between seeding and harvesting, you want all those runs done in exactly the same tracks to minimize waste.

A "small robots" revolution might still come, on the coat-tails of agrovoltaics: if/when someone in the right position begins thinking the panel scaffolding as dual use, doubling as a "rail network" for robotic tools. I believe this could become a feature of agrovoltaics installations as unremarkable as overhead cranes on factory floors.

This is the idea of controlled traffic. Use the same track over and over again to not compact the soil where the plants grow. An example of a large machine doing so is the Nexat. There are videos on youtube and some detail on their website https://www.nexat.de/controlled-traffic-farming/

Going larger there increase the percentage of growing soil vs track soil.

Reply on news.ycombinator.com