Inspiration For Weed Killing Robot Tool

Hardware
41

The reality is bindweed is like morning glory and false bindweed. None of them are slang names. All three are meant to refer to a particular species groups.
bindweed is something Convolvulus
Morning glory is something Ipomoea
False bindweed is something Calystegia
The problem here is people don’t know how to properly identify plants call something Convolvulus, Ipomoea or Calystegia any one of bindweed, false bindweed or morning glory,

All three plant types grow fairly much everywhere around the world. So yes in some areas people will say hey there is a stack of Morning glory over there and you go check and hello bindweed or hello its all 3 growing with each other.

Palomineo it been a while since I have done weed control. A general rule I forgot most cases with the three names if the person cannot give you the start to the scientific name with the name presume that don’t have a clue what they have.

Ipomoea batatas sweet potato that other name “tuberous morning glory”.
Most Morning Glorys(Ipomoea) are good edible crops. So a garden dispute between something morning glory and leaf crops can be status normal due to the fact how many morning glories are food crops.
All False Bindweed(Calystegia) are not toxic to us or farm animals but not a food crop.
Most Bindweeds(Convolvulus) are toxic to us or farm animals. Some of the bindweeds are becoming engendered species due to effective extermination.
So there is a difference between the three as a human how we should handle them. Ipomoea and Calystegia you don’t need to wear protection when removing them. Mind you the Convolvulus you should wear protection.

Bindweed is the original greek name for Morning glory, bindweed and false bindweed. It was later they were broken down into two classifications between eatable(morning glory) and not eatable(bindweed) for a long time false bindweed did not exist. Yes the first split was done by the greeks. Then we have had a lost in time were people have forgotten what they should call morning glory and bindweed. Its funny how you places like wikipedia saying bindweed is sub group of morning glory reality all three are subgroups of bindweed by history.

So I would guess there is high odds that you are using the name ,morning glory in a slang way to refer to a plant that is not morning glory by the greek rules of naming.

Ipomoea translates from greek to
The generic name is derived from the Greek words ιπς (ips) or ιπος (ipos), meaning “worm” or “bindweed,” and όμοιος (homoios), meaning “resembling”. It refers to their twining habit
From wikipedia. So Ipomoea directly means resembling bindweed.
Mind you morning glory is not the original greek it comes out of a lost in translation some how moonflower becomes morning glory in conversion to english. Then latter on with more errors people start calling bindweeds morning glory.

The biggest cause of name issue is this.
Convolvulaceae, known commonly as the bindweed or morning glory family
From wikipedia and others.
There is a problem here. If you go to the books you will find “morning glory family” does not cover all the Convolvulaceae group only the non toxic plants and bindweed does not cover the non toxic plants only the toxic ones in the Convolvulaceae group if the plant is toxic its bindweed if it non toxic it might be morning glory and that is what leads to the third group false bindweeds they are not morning glory and they are not bindweeds because they don’t meet the classification for those names. So plants that are in the Convolvulaceae group need at least 3 different common names to cover them all. Just because a plant owns to morning glory family does not mean it has the name morning glory either it could be a woodroses and other names like that. Lot of places like wikipedia have or between bindweed and morning glory the reality xor a plant is either a bindweed, false bindweed or a morning glory never really more than 1 for the correct common name. Slang names will see them with more than 1 of those names.

Sorry for such a long message explaining areas of mixed up classifications is not simple. Sorting plants into there correct name does help with garden safety at times.

42

I might be talking out my butt here but I was thinking what if they used the green detection like they are using but also implemented a pattern recognition where any thing that is green and does not lead back to the main patch of green in an area. Or any color of green that does not match the main % green in the circle.

43

Sorry I don’t know. It maybe that it is done manual. Using the camera so you can weed from in doors using your phone. Like i stated it is under development. It is hard to say untell i have a bot running. If it is not fully functional it will not be long. There should be a bunch of guys coding on this thing very soon. As FarmBots get delivered and out in the wild people will start hacking and making all kinds of cool things. This is what i find exciting about this project. It reminds me of when reprap first were built and the printers got better and less expensive. There still being improved upon.

Theoretically it should be as fast as facial recognition in cameras.

44

Hey josh thanks!

I had assumed it was further along!
That’s ok, we’re still looking into dousing bad weeds with hot water! So there’s lots of time!

45

I have one concern with boiling water. It burns. We need to keep in mind safety. Farmbot uses nema 17s because there safe. The worst that could happen is a person may get pintched. The system is not strong enough to dismember body parts. Hot water has the potential of serious bodily injury. So as you research this please keep this in mind.

46

Many parts of the video were sped up to not bore people, and the flyover weed detection was a complete simulation. Though the software that we’re developing does actually work in near real-time. So I imagine it would not take more than 5 or 10 minutes for FarmBot to go through and take the photos needed for weed detection, and then just a few seconds to process them all and identify the weeds.

1 Like
47

For the hot water delivery tool;

There should be a splash guard that would not only keep ppl safe (it would spring loaded to make sure it was against the ground, it could also work the valve so that it could not be deployed until it pushed onto something) but it would also make sure the water stayed concentrated on the target.

48

There are also instant methods that would not heat the water until right before it went out the nozzle. This would limit the incidents radius to right around the arm. The whole tool and heater could be surrounded in the shield mentioned before so there if there is a failure it would still only hit the shield.

49

Rory
That’s a reasonable time frame…not super fast but well within reason! I suspect it would be improved over the years.

50

Bagginsdada
That’s why I suggested the microwave heater… Immediately heat a 1/2 cup of water and squirt it on the weed! The addition of a guard to protect nearby people or plants is a great idea!
Tom

51

Yup that was a good one. Sounds like you could control the power of it with the same shield push switch. To simplify the programming (basically none just xyz of tool pick up and timming for time to push down) and simplify its hardware to limit fualts. If it was just a heated plate it would be less cool and probably less power efficient but almost never fail.

52

Hmmm with the potential to kill bugs with it as well!

Although!
We have foot long slugs here that can strip a plant of its leaves in just a few nights! There are black ones aliens from China or Asia… Brown ones are local and don’t usually bother my garden. To catch the black ones, If I see one I poke it with a sharp stick leaving the stick to hold it there… The next night it’s friends will come to see what all the slug screaming is about… Pull him out and use the stick to poke them all…it goes like that for several nights…
Gruesome!
The tiny white slugs here are as bad! They hide beneath the leaves though…not so easy to spot or poke!

53

The reason for using hot water is penetration so heated water to ground meets objective. Heat plate risks baking the soil into brick if heat leaks the wrong way. Yes there is so much protecting humans and there is so much protecting the ground.

I might be talking out my butt here but I was thinking what if they used the green detection like they are using but also implemented a pattern recognition where any thing that is green and does not lead back to the main patch of green in an area. Or any color of green that does not match the main % green in the circle.
There are many shade of green. Current talked about weed detection seams to be normal circle. You want to sort a weed plant from a good plant that are growing with each other then it sorting the colours.

Josh nema17 are not quite as harmless as they first appear either. Lot of nema17 are rated to operate up to 80C above ambient with a max ambient of 55C so the worst temperature a nema17 could be and still be in spec is 135C well above boiling. So the worst thing nema17 motor could do to a person a nasty burn worse than boiling water. Yes the nema17 that a human could directly touch should have some shielding. I do agree considering how to get the hot water to target safely does have to be considered but hot-water is no more dangerous than some of the parts that already make up the design. 80C above ambient on motor specs are written as “80C temperature rise” so take the ambient operating temp of motor with that and add 80 and that might be exactly what you are going to find yourself with. How people get with burnt nema17 is take them way too cheep because of their low torque thinking they are harmless.

54

Here is another interesting method: http://inhabitat.com/naturezap-uses-light-instead-of-pesticides-to-rid-your-garden-of-weeds/

55

Nature Zap DE Weed Killer is a variation of something old.done safer. Nature Zap DE Weed Killer basically take a old propane gas flame touch design like Red Dragon VT 2-23 C Weed Dragon attempted to be done using electrical means.

Nature Zap DE Weed Killer is safer and more effect than the old propane gas to a point. Then the 70-80 percent kill rate of weeds Nature Zap DE Weed Killer has is fairly much the same 70-80 percent farmbot will kill by just bashing the weed into the ground and farmers kill by ploughing weeds in. Nature Zap DE Weed Killer and propane gas touches cannot be used when there is a total fireban.

Also you will see Nature-zap DE Weed Killer has a lot of variation in reviews as well because it does not kill all weed classes.

I agree Nicholas method seams interesting but over all its not that interesting in a garden bed were bash weed into ground is option. Nature zap in past did a batch of steam product that were not effective either. Steam does not go deep into soil before losing most of it heat. Yes the Nature Zap DE Weed Killer is better than there steam products but compared against the range of weeds boiling water gets is still weak.

http://www.regional.org.au/au/asa/1998/6/315hewitt.htm
This here is a 1998 formal study of hot water vs flame vs Glyphosate. Yes Glyphosate is the product most people know as round up. Hot water applied correctly is as lethal to plants as Glyphosate with none of the chemical problems.

Key is to note what happen to flame one of the test put flame up against a weed resistant to flame and flame kill rate collapsed. The Nature Zap DE Weed Killer and the farmbot current bash weeds will suffer from the same problem as flame in the 1998 formal study fine until you get a resistant weed.

Hot water for effectiveness holds it own against the best chemical pesticides in kill rate. Yes it will take thinking how to do delivery.

This is the problem you need to look at weed killing methods with 100 percent success farmbot needs at least one of the methods. If the hot water system can be used on a wand for spraying around the garden bed it would be super great as well. Once you start looking at researched and tested methods there are not really many methods to consider. Bashing weeds is quite effective particularly when considering vs time and power. Hot water on weeds is extremely effective kill rate just not as power effective. Items like Nature Zap DE Weed Killer and flame touches using more power and time than bashing the weeds with about the same or worse effectiveness rate compared to bashing. My issue here is if you are spending more power you better be getting the results to match.

56

I meant that the heat plate would be used to heat the water near the point where the water would be pumped out to the ground.

57

How much water would be needed per injection?
This would be the first question I would need answered. If you guys think 25 ml this would be easy to heat. But if it is 500 ml this is different. The heater would not work well on the tool head. But remotely mounted. I like the toolhead manual idea. You would not even need to code anything. Just drop the weed killer.

As far as safety goes burning is only one side of the saftey. Boiling water is explosive. There is a lot of energy there. A bad burn from a stepper motor is nothing. You will not die. Exploding water tank. Maybe, maybe not. As I said safety. I am not saying don’t do it just keep saftey as a top priority.

58

In another thread there has been an idea for a above mounted rail system. If it had wireless data and track power I would then vote for more than one of the rails and multiple arms for the farmbot. This ties in here because I would then vote for the manual tool to kill weeds since there would be multiple arms and no need to worry about spending too much time on weed killing. If you could hit a weed every time it started to sprout green you would in mho run it out of reserves and kill it off no mater what the breed.

59

Josh volume of water heated and needing to be applied is linked to depth. 25ml would get weeds that grow in the top 2 inches. Going after the pest weeds under 4 inchs 200ml to 1l. It could be needle injected in 25ml pulses. Basically needle on tool head able to go in to ground up to 1 foot to deliver heat though the soil profile as it the heat that kills.

Josh yes using boiling water effectively might take a few different tool designs and will require field testing to work out what one works well. I do agree with you Josh that safety has to be considered like making sure you don’t create a closed pressure container. Steam is a double effect. As you make steam it can create pressure and create exposition if pressure build up is able to happen. Now the other issue with steam is when it cools it can create a implosion.

If you could hit a weed every time it started to sprout green you would in mho run it out of reserves and kill it off no mater what the breed.
The key word in this line bagginsdada is If. Problem with the plants that are tough that are coming up from +4 inchs down the next point they come up can be quite a bit sided ways to the bit you bashed off so come up too many times it will find it way under your wanted plants this is why you cannot do too many bash cycles.

Also there are some weed we get here where bashing them is totally pointless and they are not going to run out of reserves because they are vampires of the plant world.
http://agriculture.vic.gov.au/agriculture/pests-diseases-and-weeds/weeds/a-z-of-weeds/dodder
The worst nightmare for current farmbot detection and disposal I would think is parasitic weed dodder. This plant is a vampire no green colour to it as it gets all is food by stealing it from plants. It does not even bother having it own roots and leaves.

If you detected dodder sprouting anywhere it would be fairly much don’t bash drop heat on it and hopefully kill it before it can reach your plants. Worse is you in lots of cases be willing to terminate percentage of crop plants to prevent dodder getting any further. Yes this would be using yellow detection not green. This brings up another serous problem weeds come in a nice broad spectrum of colours. Yes dodder going after roots is pointless scorching the top part of your plant to death and hope it recovers from roots is a control method for dodder.

1 Like
60

That is a great point that the weeds can come in different ranges of color. What would be great to have is a trainable system. If we could train the image processing for different kinds of weeds. I have seen Computer vision systems that are able to detect and identify humans based on facial features. Also other people have made id systems to identify animals. I am sure this is also happening with the plant kingdom as well. Pulling from spectrometry and molecule fingerprints. Could we not come up with an opensource weed/ plant Id Data base? With OpenFarm Could this also be incorporated? If there were hundreds of people training there systems with there local weeds we could build a system that could be very powerful and beneficial. I know nothing about big data. Maybe there are data bases already with images that could be used to train the vision system. A needle injections system is an interesting concept as well. Great ideas oiaohm.