1 - there are no straight lines on a farm. Maybe in a hydro setup built on concrete slab, in which case, why robot?
2 - needs a way to cover more ground. If it's not able to travel linearly indefinitely in one direction along a bed, there's no way to scale.
3 - grit. Precision CnC mechanisms and dirt do not mix. You can't keep farm things clean (unless hydro, and even then, things get wet and caked with fertilizer salt)
Ditch the linear ways for something like bike tires. There's an open source solar farm robot which rolls around weeding (can't find at the moment, maybe was a video). Not limited by a box. They need robust mechanisms which can stand up to farm abuse and are easy to service, grease, and replace (unless you hermetically seal the components, which is harder than it sounds).
Weeds are actually pretty easy to manage.
If someone wants to make an impact in the farm tech space, come up with a cheaper alternative for scooping and dumping dirt. A ride-on tractor can be had used for $500-800. But as soon as you start talking loaders, it's $2-3000 used, and tens of thousands for new. Also the smallest tend to be a 4 foot wide bucket and a few hundred pounds lift. I want something half or a quarter of that. Able to scoop 50 lbs of dirt, repeatedly, and dig holes/trenches. That would have a massive impact in agrarian communities around the world.
As someone who went through endurance sports, being the fix-it guy, and a CS program I always sort of had one foot in both worlds and didn't really engage in this kind of rhetoric.
But as I've been accumulating more woodworking tools and getting a lot of dirt under my nails building a giant garden, I'm starting to see their point.
This device is going to gum up, being exposed to UV all day, dust, wind, and rain. The tolerances on this device are way tighter than any in situ environment is going to allow. And you're going to kneel on that track and it's gonna hurt like a motherfucker.
This will be like a pet turtle you have to keep flipping right side up, not like a roomba.
Having recently bought (and returned) the flagship Roomba, I struggle to see the difference between those two.
You say that like this device hasn't been around for years, accumulating revisions, upgrades, and user stories.
As a small hipster raised bed farmer (or rather, gardener), I actually looked into this at one point - the $1500 price tag put me off of it really quick. There's no way that kind of investment makes sense unless you're planning to sell the vegetables to recoup your investment - and it's doubtful that the machine will realistically last more than five years or so, so you'll have to replace it at some point in the fairly near future.
iRobot founders' new thing. Small, and undercapitalized, but promising.
I think it might have been this one, Acorn by Twisted Fields:
https://community.twistedfields.com/t/introducing-acorn-a-pr...
I'm not sure what you mean, to me this sounds like "everybody knows that birds can't fly". But I might just be totally misunderstanding you.
Do you mean "there are no straight lines on farms near where I live"? Cause, well, half my relatives are farmers and their farms are very much made up of straight lines.
For context, here's a google image search for Dutch potato fields: https://www.google.com/search?q=aardappelveld&source=lnms&tb...
Large, flat, level areas aren't that common, while major-to-minor deviations from flat-and-level are very common. The grandparent comment is I believe rightly asking "how does this 'farm bot' solve the problems of real farms?" I agree, this looks less like a farm-bot and more like a "serious gardener bot", which is a different niche.
[1] - https://commons.wikimedia.org/wiki/File:Palouse_hills_northe...
https://www.toro.com/en/professional-contractor/compact-util...
Here’s a nice overview of a no-dig farm: https://youtu.be/u79tiVcj8bY
The simple AB line served the agricultural guidance industry (think GPS) for many years before systems capable of handling complex curves arrived on the market.
Also their approach is great if you have a metal shop and lots of raw materials and components ready to go, but for their goal of helping mechanized developing countries, I think it falls flat. The first rule of pricing hardware is economy of scale wins. Being able to fabricobble a farm tractor from a Toyota pickup is way more valuable than building one from scratch using steel box beams.
I think their approach would be better served by creating an "API standard" for mechanical interfaces, like an ISO standard of sorts, so you could fabricobble your prime mover, your bucket loader, your hydraulics, from whatever you happen to have, and be able to swap out with anyone else's peripheral. Like the IDDS, but more down to Earth :) :
https://en.wikipedia.org/wiki/International_Docking_System_S...
https://en.wikipedia.org/wiki/Spidercam
Where it can be suspended above the field and moved to any point above the plant and can be raised as crops grow?
I do believe this won't necessarily save someone time. It probably just turns two hours of vegetable gardening per week into one hour of vegetable gardening plus one hour of robot gardening. I believe there are enough people in the world who enjoy gardening their mechatronics for this to have a respectable future, but I agree that it will never compete with any commercial venture
... unless someone invents an OpenCV saffron stigma identifier.
Even better: a slug-hunting robot powered by fermented slugs:
http://www.cs.cmu.edu/~eli/tmp/slugbot.html
The project is from circa-2000, I remember being impressed by the idea at the time, and I'd have thought slug-hunting robots would be more of a common thing by now - turns out robotics for agriculture are hard.
Isn’t that called a duck?
your slugs now shoot lasers
Props for trying though.
I’ve had a similar idea before, for roaches, but I’m just not sure how it could work. Surely any laser with enough power output to kill something would cause a decent amount of collateral as well?
I hate mosquitos as much as slugs - but I do not like people to loose their eyesight over it.
There would have to be many reliable savety mechanism, before this can become a thing.
Otherwise I will have trained one, to target ticks as well.
That way you'd be performing target recognition from multiple angles and mis-classifications would tend to only result in the not-slug being hit with 10-20% power.
Btw, did I see the robot harvesting in the end? Everything else is fine, but at least this don't take away from me! :>
I don't like this. It feels way too "Juicero" to me. Too cute, to inconsequential. It feels likely one would spend more time on building and maintaining this thing than work it actually saves and a significant amount the benefit of having a garden is working it with you hands and watching that work grow in fruition.
Bots in the farm though, they are the real deal. Reducing herbicide use by 80% or more, allowing the use of non-GMO crop while maintaining strong weed control and healthy crop, treating and optimizing every individual plant in a field, all things that will happen in the next 10 years due to robotics in the farm.
I would say that the precision part of Ag robotics is the most important part and the success of this project is likely going to be resting on that entirely.
It boils down to something like:
1. Can you sense and classify (i.e. Deep learning based vision)
2. After that, can you turn a pixel into field coordinates. (Pose-> gps/encoder/imu/etc)
3. Can you act accurately/quick/effectively enough (kill/treat mechanism, electromechnical design)
Then it's can you do it reliably enough. Then it's can you do it fast enough. Then its can you do it cost effectively enough.
Farmers won't mess with anything not reliable. They generally aren't interested in anything with worse performance than existing processes, even if greener/cheaper (unless they are significantly so). They most especially will not mess with anything that doesn't generate value in their eyes and they are very good at understanding what generates value and what doesn't.
Perhaps cooled coils: https://en.wikipedia.org/wiki/Atmospheric_water_generator
Wind power to shake the water off
Obligatory: Any opinions I express on HN are my own and not representative of my employer.
I think a few commenters are missing the point of this project. It's not a commercial robot and it's not meant to mass produce food. It seems maybe the use the word "farm" is being taken too literally. Their four stated applications are:
1. Education
2. Home use
3. Research
4. Accessibility
This is the description from their white paper [0]:
> The vision of this project is to create an open and accessible technology aiding everyone to grow food and to grow food for everyone. The mission is to grow a community that produces free and open-source hardware plans, software, data, and documentation enabling everyone to build and operate a farming machine.
Anyone who has done it knows if you have irrigation sorted there is very little needed for seedlings to grow to full size. Even with no irrigation we’re only talking 5mins watering on dry days.
Adding all this electronics and metal to an otherwise organic and natural process is environmentally, economically and spiritually detrimental.
Metals (although in tiny portions) are part of ourselves but plastics are not.
Planting seeds in a bed that size is at most an hour or two of work. Raised beds don't suffer from that many weeds, so weeding is a couple of hours per week at most. And drip irrigation with a timer can be had for a few dollars.
How much produce is produced?
How much time spent cleaning, repairing, or otherwise keeping it running
What preventive maintenance is required?
i.e., other than a learning tool, if you just consider the small garden users, what's the return?
If this robot allows people with little time to grow vegetables in their backyard, you can argue that they will move closer to soil rather than further...
The reason people don’t do it is that commercial farming is so efficient once you factor in costs growing your own is not cheaper than buying from supermarket.
I suspect this varies by location and demographics. Being an older guy living in rural areas, almost everyone I know grows at least some veggies and fruit. Not everyone makes it a major part of their lives, but I'd be hard-pressed to find almost anyone I know who doesn't have at least a couple tomato plants, or some berry bushes.
Avoid overhead watering with automatic sprinkler systems. Those systems are designed for large areas (like lawns) that need a broad application of water, not your Square Foot Garden that’s designed to take up little space. The overhead spray never gets to the root zone beneath your plants’ leaves, so the watering winds up being insufficient. That overhead spray also quickly evaporates, leading to water waste, and leaves foliage wet which can lead to pest and disease issues.
https://squarefootgardening.org/2020/05/watering-methods-for...
I do this kind of small consulting (firmware/hardware design & integration, etc.) on the side from my day job. I find that the biggest problem is discovery. There are many individuals and small businesses out there that need simple control systems built, but the only firms they can find are the ones with lots of marketing but they're too big to care about a 20-hour project.
If your friend still has this need, my contact info is in my profile. I'm in the US and happy to help.
"My dad, who I copied on this email had a quick question regarding hiring a programmer to write a small program for a farm equipment control system" is what I responded to.
They would be better served focusing on this as an accessibility device, allowing people to garden who do not have the physical capability to otherwise do so.
But if Homelab + Dirt = awesome to you, more power to you.
Some of the complaints about linear garden beds is just not valid. Different crops can definitely be linear, even grid-like. This idea can be scaled up with increasing complexity as needed.
It will be interesting to see where this kind of idea goes as the projects themselves itself grow up.
I really can't see the value add from adding a CNC machine to each raised bed.
Drip irrigation for watering and fertilizing comes to mind, which really is a better solution and prevents the need to water directly onto the leaves of the plant, which is a huge no-no to a lot of plants.
It would need a position and feedback system outside of the current "servo and home position" method, because even with closed loop motors, the distance traversed isn't exact. We need a cheap, easy way to get accurate-ish absolute position (1-3cm), maybe acoustics and phase-based?
The smallest kit is $1500 USD.
The watering seems like the actual place where time could be saved but that could be accomplished with a basic automatic sprinkler.
They even give away their own game:
> or simply impress your friends and neighbors with a quick demo
I'd love something like this, and with a refillable water tank.
Cool stuff.
* sudo apt install farmbot
* farmbot --sew strawberries
* farmbot --pick straweberriesAlso there's no scrollbar at all, that's super silly.