Farm-scale farmbot ideas

(Not sure which category this belongs in, so I’m putting it here.)

I have been pondering ways to scale the Farmbot up so it can handle something closer to the size of a real farm.

Let me be clear on what I mean. I am NOT talking about 1000 acres. I’m not even talking about 1 acre.

I am talking about half an acre. If the farmer’s markets near you are anything like the ones near me, a decent percentage of those sellers probably grow on fields of half an acre or less.

There are tractors that can handle the big 1000-acre fields. But there’s not really much out there meant for mid-range fields, where it’s too big for one person to do everything by hand, but not big enough to justify a giant tractor.

On my farm, there’s the added complication of muscle and joint problems. Makes things interesting, that’s for sure!

The good thing about my land is that it almost never needs watered. The Farmbot I would need doesn’t have to have an irrigation function. In fact, if all it did was handle the weeding, that would eliminate about 90% of the day-to-day labor. It’s too windy here for mulch or fabric, and the weeds are downright aggressive! I’m happy to do the planting and harvesting myself, but I just can’t keep pulling weeds like I need to.

So, ideas!

I’m thinking a wheeled bot, but with a guide of some kind to keep it within the lines. Something cheaper and simpler than the fancy aluminum tracks. It could be as simple as a strand of fence wire on each side.

Layout would be such that the bot can make a u-turn at the end of each block, taking it to the beginning of the next one, in a path designed to cover the whole field and bring the bot back to the starting point again. (I don’t have a sketch handy, but if you’ve ever played the old computer game “Snake”, you can probably picture it.)

The use of wheels instead of tracks means that it would probably have a hard time keeping track of it’s exact location within the block. Solution would be to have it ignore the individual plants, and just focus on the rows. Keeping the weeds down between the rows is still a huge job! There could be something at the beginning of each block that tells it what the row spacing is. Weeds within the rows would still have to be pulled by hand, but that would still be easier than weeding the whole field.

I’m not sure the “weed-punch” would be the best weed-control attachment for this job. I’m sure it’s great for soft garden beds, but not for large areas with the risk of erosion. I’m thinking a small blade, like a lawnmower blade, but tiny. 3-inches is a good size. It could even have multiple arms with that attachment, so it can go between multiple rows per pass.

If it could go through the entire field once a week, and mow the weeds between rows so they are 2" or shorter, that would be a huge, huge, help! I think that would eliminate about 75% of the work-hours I’m putting in, and 99% of the muscle-aches.

There were actually a lot more ideas I had, but they all seem to vanish when I’m at the keyboard.

Anyone have any thoughts? Questions? Concerns? Etc? Feel free to tell me how crazy I am.

This is a fun topic! I’ve also thought a bit about different strategies for automation in market gardening.

Initial thoughts on your design:

  • Wheels rolling on dirt introduces many, many more issues to deal with. Mud, rocks, roots, tools, soil, etc., make this a perception and navigation challenge that really ups the difficulty.

  • Do you mean a wheeled farmbot style gantry? I think this could be helpful for weeding, but on the other hand it would be really nice if the bot could do more manual labor than that, like harvest.

A couple ideas that come to mind…

Just use wheels on dirt just for the X axis, and maybe connect a taut guide wire that runs the length of each “track” to keep it moving smoothly. Then have a really long Y axis. Maybe even 40 feet long, or something like that. Every other row or so it could have a support post and wheel.

If the Y axis is built strong enough and you use oversized motors (wonder if you could use pipe as the backbone?) then you could hang a basket off the cross slide in such a way that a gripper could grab stuff and drop it in the basket.

It would be great if you could share a diagram of your design.

There was a piece of business advice I got . . . well, ok, it was a line from a “Last Man Standing” episode, but it still applies!

“Don’t hire people to do the stuff you love. Hire people to do the stuff you hate.”

Yes, it would be cool if the robot could handle harvesting. But harvesting is the FUN part! I don’t want to automate the part that I love best. I’d rather focus on automating the part that I hate. And since the majority of work-hours are focused on the part that I hate most, weeding, I think it makes more sense to get a working prototype that can handle that part, before adding more features to it.

Don’t get me wrong, if you don’t like the harvesting part and would rather automate that, go for it! But it’s kind of a low priority for me. Weed-control is the highest priority.


The wheeled-gantry shape looks pretty close to what I was picturing. The wheels would need to be hefty enough to handle the terrain. I’m thinking 8" lawnmower wheels should work.

40 feet long would not work in my field. I was thinking the bot itself would be 5 feet wide, with the center 3 feet being the growing area, and 1ft at each end being left for the wheels and walking space. There would be sensors on each side near the wheels, and some kind of guide for the sensors to follow. The simplest guide would be a strand of fence wire. Those are cheap, easy to install, and easy to arrange in whatever pattern is needed. The guide would need to be easy to work with, because certain crops would need blocked off when they get too big for the bot to handle. (Corn would get too tall. Squash and melons would get too viney. Etc.)

By putting the guide and sensors on both sides, the bot should be able to handle course corrections after encountering rocks and molehills and such. It just needs to stay more-or-less equidistant from the two sides.

Ellendra,
Mowing is a last resort. I have been there when the garden gets away from you because of weather or work. Roots are established and they are using the soil resources and water besides getting in the way for harvest.
Cultivating between the rows I do with the rototiller. On the slightly larger scale, is row cultivators sold with every tractor 70-100 years ago. Walk behind garden tractors like Gravely had row cultivators.
Farmbot was the first that i have seen a “weed-punch”. It is necessary on the farmbot because of lack of lateral force. The idea of scanning the field daily for weeds before they could be big enough to be “pulled” is good. My father in law liked to weed before the weeds could get established. He used a narrow rake with metal fingers to “weed” around the plants.
The first gardenbot that i have seen was in Austria at a museum/exhibition. It was circular about 2m across with a industrial arm mounted in the center. The visitor/operator could plant seeds, water, but I do not remember “weeding”? That was 25 years ago. I have wanted to build a garden robot since then. It has taken many form in my mind. At first it was a center pivot with antenna tower out to an A frame with 2 wheels with motors that would make it travel in a circle. The horizontal 3 legged antenna tower would have a trolley with small rototiller device that could till or weed or water.
I have gone through other iterations. A gantry as wide as the garden strip(50ft) with similar equipment. A 6 legged robot that could step over the vines of the cucumbers and squash. a 4 wheeled robot with two arms, one to push the garden plant aside so the vision identication could see the weeds and the other arm with a side grinder rototiller ( wine industry calls them "badgers) that flexes in multiple axis .
All that dreaming and here I am years later with my rototiller and hoe.

1 Like