Automated Grassland-weed Spot-Sprayer

[agitechi]

Hello all,

Several years ago I wrote a thread asking for advice and feedback for my then, MSc project (thefarmingforum.co.uk/grassland-weeds-sprayer-technology.296540/h). I received interesting and valued feedback and would like to update about our progress since.

I'm pleased to report that after a lot of work, development and testing is going very well and has progressed to field trials.
Here is a video from ongoing spring testing:

Without doubt, a highlight of these trials has been the data that it produces.

• Weed populations (average ground cover %) of 18% across farm. Ranges from 40% to 7%. (Mainly Docks). (Has there been data before that actually quantifies the populations?)
• 77% reduction in chemical quantity for sprayed area. 0.45L/Ha equivalent conventional rate.
• £47/Ha saving in chemical cost (Forefront @£30/L). Grass and Clover haven't suffered also.

The technology uses cameras, computers, trained "AI" software and nozzle valves to "See + Spray".
It detects and targets mainly Docks, but also identifies and sprays other common weeds, Thistles, Nettles, Dandelions etc.

As always, all and any feedback is welcome

Colin

 

[Jdunn55]

When would you be hoping for these to hit the shelves as such? Would it be in the shape of kits that you can adapt onto existing sprayers or is it likely to be an entire new sprayer needed? Dare I ask how much you would be talking as a ballpark figure or is that yet to be decided?

Finally, well done! Exactly what's needed in agriculture with the spotlight on environmental issues!

 

[Jdunn55]

Just watched the video as it didnt show up at first would running the boom closer to the ground with more nozzles reduce chemical usage further and also mean that less grass and clovers would be sprayed?

 

[neilo]

Hello all,

Several years ago I wrote a thread asking for advice and feedback for my then, MSc project (thefarmingforum.co.uk/grassland-weeds-sprayer-technology.296540/h). I received interesting and valued feedback and would like to update about our progress since.

I'm pleased to report that after a lot of work, development and testing is going very well and has progressed to field trials.
Here is a video from ongoing spring testing:

Without doubt, a highlight of these trials has been the data that it produces.

• Weed populations (average ground cover %) of 18% across farm. Ranges from 40% to 7%. (Mainly Docks). (Has there been data before that actually quantifies the populations?)
• 77% reduction in chemical quantity for sprayed area. 0.45L/Ha equivalent conventional rate.
• £47/Ha saving in chemical cost (Forefront @£30/L). Grass and Clover haven't suffered also.

The technology uses cameras, computers, trained "AI" software and nozzle valves to "See + Spray".
It detects and targets mainly Docks, but also identifies and sprays other common weeds, Thistles, Nettles, Dandelions etc.

As always, all and any feedback is welcome

Colin

That is impressive, well done.

If you can bring it to market at a reasonable cost, ideally as a retrofitting kit, you’ll be onto a winner.

In the meantime, if you want another farm to try it on.....

 

[agitechi]

When would you be hoping for these to hit the shelves as such? Would it be in the shape of kits that you can adapt onto existing sprayers or is it likely to be an entire new sprayer needed? Dare I ask how much you would be talking as a ballpark figure or is that yet to be decided?

Finally, well done! Exactly what's needed in agriculture with the spotlight on environmental issues!

Big questions, which I don't have all the answers for yet unfortunately.
As a system, it is quite far developed and needs a little more tweaking and improvements and then to be packaged up and marketed.
Perhaps there are people, companies out there that would be able to help with this next stage?
Keen to hear

 

[agitechi]

Just watched the video as it didnt show up at first would running the boom closer to the ground with more nozzles reduce chemical usage further and also mean that less grass and clovers would be sprayed?

Yes, absolutely it would. There are some systems abroad, for other crops, that have hundreds of valves per linear metre and can be very precise.
Luckily the selective allows for inaccuracy, but an approach as you suggest could, in theory, use a total herbicide (roundup), so long as they're available.

 

[agitechi]

That is impressive, well done.

If you can bring it to market at a reasonable cost, ideally as a retrofitting kit, you’ll be onto a winner.

In the meantime, if you want another farm to try it on.....

Thanks!

We'll add you to the list

 

[hollister]

Looks great, moving forward it needs to be combined with mix at use technology so you only have to fill the tank with water and then only use as much chem as required. As it stands how do you know how much to put in?

 

[agitechi]

Looks great, moving forward it needs to be combined with mix at use technology so you only have to fill the tank with water and then only use as much chem as required. As it stands how do you know how much to put in?

Good point, direct injection probably is the way to go.
I wanted to have a go at bodging up a system and link it direct to computers for chem pump activation. However went with recirculating system for this prototype

For the trials, am using the the data (average chem L/ha etc), combined with pre-walking fields to judge how much to put in. Am filling with smaller amounts (still at label rate/dilution) but am always accurate enough with the data and a guess based upon a walk of the field.
Some tanks seems to never empty but there's a surprising variation between, and within fields that eventually empties it out.

 

[neilo]

Good point, direct injection probably is the way to go.
I wanted to have a go at bodging up a system and link it direct to computers for chem pump activation. However went with recirculating system for this prototype

For the trials, am using the the data (average chem L/ha etc), combined with pre-walking fields to judge how much to put in. Am filling with smaller amounts (still at label rate/dilution) but am always accurate enough with the data and a guess based upon a walk of the field.
Some tanks seems to never empty but there's a surprising variation between, and within fields that eventually empties it out.

While direct injection, or more valves/nozzles per metre would reduce chemical use further, they would add to the capital cost considerably, reducing your sales.

If you can keep the cost down, a lot of farmers would be interested I suspect, but if it becomes a more complex/expensive system, then you will only sell a small number to contractors.

 

[Jdunn55]

While direct injection, or more valves/nozzles per metre would reduce chemical use further, they would add to the capital cost considerably, reducing your sales.

If you can keep the cost down, a lot of farmers would be interested I suspect, but if it becomes a more complex/expensive system, then you will only sell a small number to contractors.

Whilst I agree with what you're saying, for me the selling point is that you could tackle docks (and possibly more like thistles in the future?) in leys containing clover (and again possibly more like chicory and plantain etc in the future? @agitechi ). When I watched the video I felt like it was still spraying a lot of grass which in a clover rich ley would also mean you're reducing clover.

Maybe an option would be to have 2-3 different models with different amounts of nozzles? One with loads for contractors, one with a reduced amount for smaller farmers and maybe one in the middle targeted at larger farmers but not quite contractors?

 

[agitechi]

While direct injection, or more valves/nozzles per metre would reduce chemical use further, they would add to the capital cost considerably, reducing your sales.

If you can keep the cost down, a lot of farmers would be interested I suspect, but if it becomes a more complex/expensive system, then you will only sell a small number to contractors.

Agreed, ideally a commercial kit, or system would be as close to a normal sprayer system as possible.

 

[agitechi]

Whilst I agree with what you're saying, for me the selling point is that you could tackle docks (and possibly more like thistles in the future?) in leys containing clover (and again possibly more like chicory and plantain etc in the future? @agitechi ). When I watched the video I felt like it was still spraying a lot of grass which in a clover rich ley would also mean you're reducing clover.

Maybe an option would be to have 2-3 different models with different amounts of nozzles? One with loads for contractors, one with a reduced amount for smaller farmers and maybe one in the middle targeted at larger farmers but not quite contractors?

Yes, these software systems can be trained to spot whatever you want. Different crops and weeds are simply a matter of training the system and fine-tuning as you go.

There are adjustments to be made on the response time and the duration of the valve ON time. The latter is primarily based upon forward speed. The duration was longer than ideal for testing but is easily adjusted down. These adjustments would significantly reduce any overspray.

I take your point though

 

[Yale]

Love it,would really suit our grass based system.

Just make sure the electronics are simple and robust.

 

[Jdunn55]

Yes, these software systems can be trained to spot whatever you want. Different crops and weeds are simply a matter of training the system and fine-tuning as you go.

There are adjustments to be made on the response time and the duration of the valve ON time. The latter is primarily based upon forward speed. The duration was longer than ideal for testing but is easily adjusted down. These adjustments would significantly reduce any overspray.

I take your point though

I see, if the overspray could be reduced to a decent level then there would be no need for more nozzles I guess! That's good to know about the weed spotting being attainable, do you think it would be able to tell the difference between chicory and docks which are very similar looking?

 

[Formatted]

Well done, imagine a lot of work has gone into that.

 

[theboytheboy]

Worth speaking with the small robot guys?
I was very interested in what they are doing but they seem focused on arable at the moment.

Better you you may be able to do something really good for grassland

 

[Countryboy]

My son has just developed and built this automatic spot sprayer based around a robotised McConnel Robocut for his Harper Adams dissertation -

 

[agitechi]

Well done, imagine a lot of work has gone into that.

Thanks. There really has been, and especially from the help of brains much bigger than mine!

 

[agitechi]

My son has just developed and built this automatic spot sprayer based around a robotised McConnel Robocut for his Harper Adams dissertation -

That's awesome