We are delighted to see The Weedinator as an entry for the 2018 Hackaday Prize! Innovations in agriculture are great opportunities to build something to improve our world. [TegwynTwmffat]’s Weedinator is an autonomous, electric platform aimed at small farms to take care of cultivating, tilling, and weeding seedbeds. The cost of this kind of labor can push smaller farms out of sustainability if it has to be done by people.
Greater efficiency in agriculture is traditionally all about multiplying the work a single person can do, and usually takes the form or bigger and heavier equipment that can do more at once and in less time. But with an autonomous robotic platform, the robot doesn’t get tired or bored so it doesn’t matter if the smaller platform needs to make multiple passes to cover a field or accomplish a task. In fact, smaller often means more maneuverable, more manageable, and more energy-efficient when it comes to a small farm.
The Original Weedinator was a contender for the 2017 Hackaday Prize and we’re deeply excited to see it return with an updated design and new people joining their team for 2018. Remember, there’s money set aside to help bootstrap promising concepts and all you really need to get started is an idea, an image, and documentation. There’s no better opportunity to dust off that idea and see if it has legs.
42 thoughts on “The Weedinator Returns”
Man, those little Honda generators are hard to beat. Will be interesting to see the implement attachments as well as sensing methods. Seems data on the field characteristics for watering, fertilizing and nuisance invasion can be gleaned while operating also.
The first implement I’m going to test is a rotating claw on a CNC platform. Some of the others in the team are going to be using 3 point links and some other interesting devices.
Should be able to start working with object recognition cameras fairly soon but need to build the outer canopy to control the lighting against glare from the sun. I’m trying not to get too carried away with getting into the exciting stuff and instead spend some quality time getting what I’ve got so far working really well – castles built on sand and all that!
Very neat project
Curious what happens when (if?) the bar gets raised and every small farm needs to have this (or larger equipment) to be able to compete?
“it doesn’t matter if the smaller platform needs to make multiple passes to cover a field or accomplish a task”
It does if you need to harvest everything RIGHT NOW because, well, it’s harvesting time or else it rots in the field and your entire year’s crop is worthless.
Then drop the farming subdisies that mostly benefit the larger industrial farms.
Price of food goes up, but taxes go down. Plus minus zero.
Yeah, more costly food. Less money left for geeky stuff. I can see this crowd getting behind that.
You forgot the part where you pay less taxes, leaving you just as much money as before.
Oh, who am I kidding. Taxes never go down.
-80% farming subsides, +0..05% military & defense budget
Though the price of farming really doesn’t have anything to do with the price of food, which is 95% up to the processing and retail chain.
Ask any farmer what they get for a pound of beef, or enough grain to bake a bread. Then compare what it costs in the supermarket.
More precisely, the cost of the actual food in your food is 14% and the other 86% is processing and retail cost.
Ie. food costs a lot because it passes through many hands, each demanding profit on the transaction, plus he government demanding taxes.
Thanks for the positivity! It’s more about small farms competing with big farms although obviously they will compete with each other as well to a certain extent. I imagine these machines would be more ‘modular’ than the big tractor on the big farm scenario so rather than having one $100,000 machine you might have 10 x $10,000 machines.
That was one of the most disappointing project links I’ve ever clicked on here.
I went to the project and found the project description written like a marketing blurb that was insistent on not revealing what you’d actually be getting. Or maybe they haven’t thought it through yet.
What’s presented so far, looks like a mobile bed with GPS. Nothing on what a farmer would have to do to use it, it if its wheel spread is adjustable to different crops, wheel height or bed height for different height crops or crops at different stages of growth, how the weeding will actually take place (leave soil vulnerable to wind erosion? etc.), what area it may be able to process for a given time or anticipated energy use.
Has anyone talked to anyone who uses any of the hardware, for example lead screws, in a dirty environment? I’m reminded of the sage advice passed to me of “Never buy a used combine from a bean farmer!”
In weather? I can hear the farmer now “Well the weather’s too much for the ‘robot’, I’m going out to the fields to get the work done”.
Perhaps if the link to the “original” shows the original project instead of the 2018, maybe it showed what they’re trying to do and how.
Looks like a bunch of people playing with some gear that may turn out a bed that someone might be able to turn into something useful. If it could handle the range, slope and soil type, maybe it could bring the farmer out their lunch.
Sorry you’re disappointed in my project – I’m trying really hard to make some significant progress with it and …. as it says in the description …… the logs will be updated as and when. Please subscribe and be patient!
At present, the machine will only be doing one simple task – weeding – which involves disturbing the soil around the crop before the tiny weeds come to the surface. It’s more like weed prevention than trying to pin point individual weeds like the Farmbot. The machine will use a rotary claw as described in WEEDINATOR 2017 which is moved on a CNC type platform.
I’m sure that there will be all kinds of interesting problems that become revealed as the machine actually starts work in the late Spring, abrasion by soil particles being one of them. I’m sure it will do more than just bring out my lunch!
“At present, the machine will only be doing one simple task – weeding – which involves disturbing the soil around the crop before the tiny weeds come to the surface. It’s more like weed prevention than trying to pin point individual weeds like the Farmbot. The machine will use a rotary claw as described in WEEDINATOR 2017 which is moved on a CNC type platform.”
Small ‘Roomba’ style machines out already. Instead of digging down into the soil, they simply skim the surface and chop off the tops of the
tiny weeds before they get established. A lot less power used (no digging/rooting). Mobile bed doesn’t need to be able to stop in place to let a heavy rotor be forced/dig through the ground. Soil largely undisturbed; retains moisture. No dirt/dust thrown around.
Very strongly suggest you change to a modular approach to project: the ‘autonomous’ bed that moves around the field; attachments that quickly & easily change out to sit on it and do work. Farmers love attachments that make something multipurpose.
One thing that is needed is directed application of fertilizer (organic or otherwise), instead of spraying the whole field. That bed should be able to take an attachment that will apply liquid or pellet fertilizer at the base of the line of the plants; or to the soil between rows, depending on what a crop requires.
Same for watering, without having to spray water up into the air and have a portion evaporate away before it even hits the ground.
Once you’re dropping liquid onto the ground from a very low height (instead of spraying it), it’s very easy to have a metered flow rate.
Once you can determine attachment weight and generator/battery-charge providing what range it can service, you may be able to have it weed, fertilize and water in a single pass down a row.
With your attachment delivery of fertilizer or water, consider having delivery rate vary based on GPS location in field, so a large field can have custom delivery.
Need WIFI type messaging between base and mobile unit, so farmer can be alerted when attachment needs something (gas, fertilizer, water, etc..). Consider farmer going to service supplies at unit in place or it returns to the start of a row, vs. unit returns to base.
Consider what of the above it is cheaper to design/implement now instead of having to redo (nearly) everything later.
Given the size and strength of the bed, it should also be able to accommodate an attachment that forms the rows.
One that plants the seeds at the required depth and spacing. Appropriate water/fertilizer at same time, on or between seeds as appropriate.
One to even plant seedlings.
Yes – all these things are possible and no doubt I will do some of them. I’ve deliberately limited the scope of this project to weeding to make it more manageable or else it’s likely to become totally over-facing. No doubt this thing will take many years to fully develop – the opportunities are almost endless!
PS. Thanks for the tip about the gaiters / bellows for the lead screw – I’ll get onto it.
The point is, make it setup for modules from the beginning. Make the physical bed independent of the add-ons. Your weeder should be able to be removed/installed from/to the bed in five minutes or less.
Less total development work.
Even easier to swap out/in a new weeding design should you decide that’s necessary, as there is likely to be little if any modifications to the bed.
re., gaitors/bellows: and wipes/blades, for when the bellows leak… google.
Don’t forget shields to protect the bellows from flying dirt. Cost of maintenance and downtime.
These things are cheap compared to down time to clean lead-screws or linear-bearings, and that’s if they can be cleaned and don’t need to be replaced.
Start thinking about HOW a farmer would use your ‘product(s)’, and why they’d want to use it over another product or another solution.
I just checked the photos again.
Nice linear bearing. Except in dirty environments. The bellows needs to protect the rail too.
The more expensive higher precision bearings have the most trouble with dirty environments. Just ask the people that tried to use them. For example, in wood CNCing environments where the wood fibres get thrown on the rails (even with the wipes), the shop I saw had the top/most-expensive recirculating bearing and they needed an hour of tear&clean for every hour of use.
You can machine a custom pillow block that will hold machined pieces of Delrin AF as the contact surface to slide on those rails. Not the best rails for that, but it works without all of the tear-down & cleaning time those bearings will need.
I like the idea. Seeing how it goes in use will be nice. Hydraulics have a long history of success in farm equipment. Particularly complicated and adaptable machinery like self-leveling combines and root vegetable harvesters, etc. Keeping lead screws and electric motors going is a challenge and if it works out, very impressive. I imaging this in the end is meant to be solar powered? Maybe I missed mention.
When I saw the frame and generator I wondered about a bit more power and hydraulics. China and India and SE Asia are full of those nice little diesel engines that are used for everything. Mount one of those as a power plant. (I keep thinking about getting one.)
I did think about hydraulics but went for electric due to cost and ease of control. I guess I should think about protecting the leadscrews with some sort of rubber gaiter?
It’s been done before. Don’t reinvent, but you can do your own implementation. Look at what is done for screws and linear bearings.
Also, add a physical shield to protect the containing bellows from projected dirt so you’re not repairing/replacing the bellows as often.
If you’re going to use a linear bearing in that environment, look at contaminate tolerant materials, like Delrin AF, that work when dry direct on hard steel. With dust blowing around, a tiny leak and a traditional quality lubricated screw or bearing can jam up and require a complete tear down and cleaning. If you have an issue with dust, the nut on your lead screw can also be replaced by machining and ‘annealing’ a Delrin AF replacement.
I’m seriously thinking of using a shrouded flame for weeding rather than rotating claw. Not so much dust and soil flying around.
Having done enough weeding of “small” gardens ranging from our own little raised beds to 1 acre+ of sweet potatoes or pole limas, this is not only a great idea, but a bigger than it seems reach.
Have to agree exposed lead screws probably not best idea diving and digging in mud or soil. I’m following. Be interesting to see actuator design that develops to pull weeds. 2kw Genny seems excessive. Even if nice efficient Honda.
From the comments here I’m seriously thinking of using a shrouded flame for weeding rather than rotating claw. If anything the generator is underpowered as the drive wheel motors are currently being upgraded.
How fast is this machine going to be? Was thinking less than 6fps. More to around 3fps. Is there going to be carry -with like pesticide? Flame thrower shrouded or not is likely liability. Looking forward to solutions.
If mechanical weeding, driven by AI/machine learning robots like this one, ends up being cheaper than RoundUp (about $10-$15/acre), this is going to be a no-brainer for farmers, even if the resulting crops don’t fetch a price premium because they’re organic. Imagine a solar powered version: weed all day, sleep at night, unless it can collect and store more energy during the day than it needs to weed, resulting in extended operating hours. (Or if it’s important to weed 24/7, set up a small bank of solar panels at the edge of the field, storing solar power in batteries that can be used to recharge the robot.)
It’s not just the money cost with roundup but the environmental issues. The chemicals kill all the insects on which birds feed so affects the whole ecosystem. Since the use of widescale pesticides insects have taken a massive hit – the car windshield test – 20 years ago your windshield would be covered with thousands dead bugs and now you’re lucky if you get a few.
Eventually the machine will be powered by lithium ion batteries but they’re just too expensive at the moment. Charging is fairly quick but would be probably be done off the main grid rather than solar panels, but that is still a possibility.
Round-up is an herbicide, it has very limited impact on insects at large, it’s toxic to a few species of beetle but that’s it. Even at higher than prescribed doses it doesn’t adversely affect insects.
Your car windshield test says nothing about the unnatural swarm sizes only made possible by large scale agriculture. Just because it was one way 20 years ago doesn’t mean that way was any better.
At the temperatures Roundup is to be applied, it decomposes in 24 hours. It only takes seconds to trigger the growth genes in broadleaf weeds.
That’s incorrect. It takes 90 days to a couple years to fully break down, glyphosate / roundup is even found in cow manure and some compost.
It is changed sufficiently to not kill plants in 24 hours.. You can seed new grass in 24 hours for example. Glyphosphate has many sources that produce much more than Roundup.
Grasses & cereals are different from other food crops. Lots of crops like nightshades and beans don’t tolerate very much at all.There’s also lots of dofferent concentrations sold as ’round-up’ some meant for dilution others as direct application.
Organic produce around here
costs 2 to 3 times more n the stores.
That is just about exactly the yield for organic crops. One third the yield of modern farming for the same amount of land and work. Assuming no pestilence.
What, no video of it in action. Dont post projects that have no videos that show it in action.
Oh sorry – there is a video on the page for the hackaday prize but it’s hard to find. I’ve updated the details now with a video – thanks for the tip!
any videos of it actually removing/preventing a weed (or the mechanism for that)? or is that part of it yet to be implemented?
Not yet – should have something for weeding in early May. Please subscribe to the project for updates.
looking forward to seeing this! great work..
Thanks great article. Many machines are moving to autonomous driving technology going forward. $300,000 tractors now can navigate an entire field using gps.
Please be kind and respectful to help make the comments section excellent. (Comment Policy)