Scaling FarmBot to bigger farms

Hi there,

I have seen the discussion of scaling to bigger farms is now always around bigger hardware.
Did you consider the idea to have same hardware, but longer tracks? Like you can have 200 beds of 100 m long and split them to segments and the FarmBot can handle one segment at a time. After finishing with it, to go to the next segment and so on. Probably it will need some wheels on those long tracks and when moving to next segment will just use them to fix themselves to the next segment.
Not sure if makes sense, but would be great to find a solution to fit any farm size and I would assume if the farmbot could move, then it could handle any farm size.
What do you think?


This raises the question - what percentage of the time is FarmBot active in the 1.5x3m Genesis? If it’s working an hour a day (making that up as an example) then you could only scale one FarmBot to 24x the area before it’s 100% busy, so it could just keep up with a single, 72m long track.

Or if it were only 1% busy (14 minutes a day), it could maintain 300m of track. That’s a lot of food!

1 Like

I don’t think it makes technical sense to have one FarmBot move from one bed to another. It makes much more sense to just extend the track and keep the device attached to that one set. As Laird mentioned, the main limiting factor is the number of hours in a day (assuming you solve any hardware difficulties). My FarmBot currently only operates for about 30-60 minutes a day to do watering in the morning and evening, and sometimes mid-day if its really hot out.

But, if you had your FarmBot monitoring the soil often, or detecting and removing weeds, or taking photos, or whatever else, then you could probably have it be doing stuff for 2 hours a day. In which case you should be able to scale up to 12x the size without an issue if you want to fully use the 24 hours available in a day.

But is that really the best? What if its most optimal to water within a certain time period, and then to measure soil moisture within an hour? Then you have to make sure FarmBot can tend to all the plants within a 2 or 3 hour period maybe. And then its idle the rest of the day. It all depends on what you want to do with it.


Understood! Thanks a lot for your answers!
Indeed, if the bot needs to work for 2 hours per day, than it is much more complicated. Maybe a mix of traditional tools and the bot would work…like the irigation and sensors / monitoring to work intependent of the bot and the bot to do the weeds and other human labor dependent stuff.
I am investigating the option to use farmbot for a small / medium farm of 13 ha of land, but I assume might be complicated to scale to that size for now.

1 Like

The autonomous driving robot along the lines of the Hortibot, Ladybird, or Weeding Robot from Naio technologies is likely a better platform for larger farms.

That said, I am still looking for a peer to peer/open source project for such a device! If you know of one please let me know. (Ladybird) (Hortibot) (AgBot II)


So is takes one “fully developed” acre (not to mention “resting” land) to feed one person per year. With the current configuration, it does not seem feasible or realistic to have a FarmBot (or bots?) running over an acre 24/7/365. Anything less than a fully sustained loop (so a family of 4 will need 4 acres with full automation) seems like an amateru(ish) idea. What am I missing? Are these tools capable to scale that way in the future? Do they ever break down (say harsh climate, floods, winds, power outages, idk, stuff…)??

1 Like

"So is takes one “fully developed” acre (not to mention “resting” land) to feed one person per year."
I guess the accuracy of that statement depends on climate and soil quality. My guess is that figure derives from a much less intentionally managed acre of land. Per Curtis Stone’s website, he got 50000 lbs of food off of less than an acre of land. I don’t know what that works out to in calories, but the sale of that produce provides Curtis and his family with a better than living wage, so it certainly feeds more than one person per year. I am not claiming that this system could replicate his systems, but just mentioning this to show the variability in possible production.

This is the general maths using standard farming methods. The result for everything is basically 2 acres in decent environments. But there are ways to get smaller.
Number 1 nuke the wheat and learn to make do with rice and corn.
Aquaponics and Hydroponics can cut particular crops to needing only a percentage of their land area sometimes quite extreme like lettuce what is 1/100 so for every 1 lettuce you will get in soil in the same area you can get 100 using Aquaponics and Hydroponics. Aquaponics gets to double stack fish growing and group growing so hopefully you like fish.

Producing a well round diet off of under acre of land is possible but not easy. Mixture of intensive cropping with aquapoincs would be required. Problem here is if you don’t care about well round diet you can produce way more mass of food in a acre of land.

Curtis Stone’s cropping is not design with the goal of producing a complete balanced diet instead design around produce volume and use the sold excess to fill in the gaps.

I definitely agree that intensive cropping is necessary, but that is what we are talking about here, in fact this could be a whole new level of intensive cropping . If you are going to bother to invest in this much technology it seems like an awful waste to not use it to its potential. The key to really high productivity is a combination of correct crop spacing, companion planting, crop succession and soil building. So though a standard gardener would plant a row of carrots, a row of lettuce and area of beans, a FarmBot could mix dozens of plants together to break the spread of pests. Predefine plant succession schedules so that as one plant ends its season another is just starting its own. This would all be too complicated in any normal garden but could be simple in a Farmbot garden. All you’d have to do is build a database of all the plants that work in your area and then project their size during the track of their life cycle so that complete ground cover is always maintained to nearly eliminate the possibility of weeds. (no light, no weeds) That is what the future of this tech is.
Now look at these numbers for monocropping possibilities and realize that with companion planting the individual densities of each plant goes down but the overall density of production goes up. So instead of planting an acre one vegetable, you could possibly plant only a quarter the density of that vegetable plus a quarter the density of eight other vegetables and end up doubling or tripling the overall production. The reason the math changes is due to breaking the pest cycles (i.e. the potato bug wants to eat potatoes and if they are spread out they can’t just hop to the next one. Each plant needs different micronutrients and if you mix them then they don’t have to compete with one another, also nitrogen fixing plants could be mixed in to provide nitrogen for all their neighbors to avoid the need to seasonally rotate crops or leave ground fallow. Next with increased production density you increase the effectiveness of season extending structures like cold frames, hoop houses and green houses. These then lessen the need for wasteful long term storable foods like grains, so as the technology matures, it continues to reduce the need for land area as it improves soil which also raises productivity.
Also besides overall food production, look at lowing food waste. Let’s say you eat a salad every day, which needs one head of lettuce which matures in 35 days. So normally you would just plant a bunch of lettuce and much would go to waste, but with this system you would plant one plant a day exactly 35 days before you need it and then harvest each on its maturity date for your salad. But that 35 days is really a guess, maybe it takes 40 days in the early Spring and then only 30 in Mid-summer and then 35 in the Fall. All of this could be accounted for and included in the program. So even though this may actually drop production slightly, it massively lowers waste so increases overall calories delivered.
What I’m saying is using Farmbot to its full potential is so different from standard farming it would be nearly impossible to make a fair comparison.

1 Like

RichardJHauser seasonally rotate crops with some crops is unavoidable while in soil. Tomato and sweet potato come to mind where you will end up with root rot issues if you keep on cropping in the same area.

This is where using aquaponics comes in as well for some crops. Lettuce matures in about 20-22 days in aquaponics and can be planted tighter than soil based. That is where the 100 to 1 comes form faster growing and high density.

No single tech addresses all the problems. Crop production is most like multi tech method.

That’s my idea too. Same size rows, longer rails. Perhaps every 3-5 rows another FarmBot.

According to biointensive farming practices the smallest space required to feed one person per year appears to be around 4000 square feet of growing space (in a California climate with a very minimal vegetarian diet). Add on roughly the same amount of land for pathways and compost piles and you could feed approximately 5 people per acre. This is a high estimate, so for the average gardener in a temperate climate I would estimate that 2-3 people could live off of an acre. If one farm bot could manage an acre of intensively cultivated crops then I think it’s a very worthwhile addition to the garden or farm.

Hi guys,

I’ve literally just stumbled across farmBot, looks interesting and i’d like to get one but is there any FarmBot prototype that has been scaled to a longer farmed section…say 10m x 1.5m etc… If so, where could i purchase it?

Hi Keith, welcome to the forum :slight_smile:

As far as I know there are no adaptations of FarmBot on a larger scale yet either for sale or open-source plans, though we’re working on stuff here at for release probably near the end of this year. So your options right now are to build your own device either completely from scratch or from a modified/augmented smaller kit, or wait until we (or someone else) comes up with plans/kits for purchase at the size you want.

Hi Rory,

Thanks for the reply. Yeah i’m not too clued up with the electronics and coding side so best I wait for you gents to get there…Good luck, let me know when I can order a larger size to RSA.



In actual practice, it takes less than a tenth of an acre to feed a person, on a fairly standard diet in northern climates. This is in actual practice in former Soviet nations. That would of course require growing rabbits, chickens or fish, or being vegetarian. It would also require working around fruit trees. No need for a minimal diet, there’s plenty of food in that space.

Scaling up:

The track system is inherently cumbersome. Let’s think out of the box.

A: Make the unit independent. It is an autonomous robot.
B: Instead of tracks to give it position info, it uses GPS. Standard GPS is only good to about 3 meters on a good day, but with 8 channel receivers and using raw data, you can have a base station that is fixed, but transmits it’s current location, as calculated by the current constellation of satellites. The robot receives this info and uses it to correct what the same constellation is telling it. This gives about 10 cm accuracy. If you need more info, you can pull phase information out of the carrier signal to get sub cm. accuracy. However for this to work, losing the signal even briefly requires you to go to a known location to re-acquire position.

Alternate position system:

A theodolite measures angles with accuracy as good as 0.1 minutes or 6 arc seconds. (vernier measure) If this can be done electronically, then you could place several corner reflectors around a field, and have a a rotating laser measure the angles between reflectors. 1 degree at 100 yards is about 5 foot error. 1 minute is 1/60 of this or about a an inch. This degrades as the angular difference gets smaller, but with a reflector on 4 corners, you have a lot of redundancy. This in principal allows a hectare to be managed by one unit.

For most uses this is adequate. For watering you have a problem, in that this system does not handle hoses gracefully. I don’t think coming back to fill up with water every few hundred gallons will work. However, if you zone it appropriately, you may be able to make a conventional sprinkler system work for this. Rather than bury the lines, hang them from an overhead trellis. Google “hop yard” to see what I mean.

The present system isn’t good at identifying weeds. This is especially problematic as plants expand into their surroundings. (Consider a squash with tendrils spreading over 10 feet.

System needs a camera, along with software such as “snap leaf” to identify the plant.

Just checking in to see if anyone has heard of any progress made in the past year to develop a grid-based commercial-scale agricultural robot? I have seen that much progress has been made on rovers, and of course Farmbot has released the new, awesome MAX versions, but curious if anyone has further explored the idea of using rails on an even larger plot (and perhaps elevating the rails several feet off the ground). One challenge I am working through with this idea is how to handle pest control - crop cover seems like it would get in the way of farmbot’s operations. Thanks!