Hi-Tech Weed-Killer

Makarand writes "Instead of making improvements to automatic mechanical weeders, Engineers at UC Davis have been busy developing the next generation robotic weeder which will use computerized images of crop rows to identify weeds and zap them. The system can identify weeds from the regular crop by assessing shape, color, size and other variables from the captured images of the crop row. A robotic cultivator will then blast weeds with a weedkiller using syringes mounted on a tractor. A GPS allows the system to calculate weed type densities within the field and the amounts of chemicals dispensed in the area."

Comments

Wow, what will they think of next...

[fpp]

...a robot that mows the lawn?

Weed killer?

[BortQ]

I'll bet the DEA is super happy about this.

Crop rows?

[pphrdza]

You've got to have your plants in rows for it to recognize them?

Guess that eliminates my garden...

weed zapper

[Anonymous Coward]

If it can recognize the weed, how much harder would it be to design an arm to *pull* it!?! (or roll it and smoke it :)

Re:weed zapper

[istartedi]

I was going to say that, but you beet me to it (groan).

Pulling is probably not practical. Some weeds like dandelions have long tap roots. Either the weed is very hard to pull, or you break off the tap root and the weed is back up in just a few days. Sometimes I can wiggle a dandelion juuust right and get most of the tap root, but more often than not I break them. I imagine training a robot to do it would be pretty difficult. Of course, I don't try to dig 'em because that would just make the lawn even uglier. In a field you don't care about things looking pretty, but then you'd have to worry about damaging the root system of the crop plants; so digging is out.

That said, it would be nice to have an organic farmer's version of the robot that ran more frequently and clipped the weed at the base.

NOOOO!!!

[gik]

Weed killer???
how will i get my buzz???

the ramifications of using this product are way too much for this soul to bear.

time to "get rid" of my weed before someone else does ;)

Evolution

[xMonkey]

You know virtually all systems have bugs in them.

I wonder how long until the weeds find the exploits.

Benchmark

[zaibutsu]

I remember working for a company which was trying to sell computer hardware to an organisation developing a system like this.

The system had a plant recognition benchmark we had to run. It was calibrated in 'cabbages per second'.

it has been sent from the future...

[intermodal]

to protect Sarah Conner's garden from her enemies...

I question the demographics...

[djh101010]

Around here (midwestern USA), the farmers are reluctant to try even the most basic of new things (no-till farming, organics, etc). I'm having a hard time imagining any of the farmers that I know shelling out a couple of grand on a robot with a camera, to run up & down the fields.

Then again, if they can show how the cost is offset by gains in yields, then it just might get some use. Another concern is battery life - just how far is this thing going to go on a charge? 1 mile? That'll cover 4 rows...then what?

Re:I question the demographics...

[Osty]

Around here (midwestern USA), the farmers are reluctant to try even the most basic of new things (no-till farming, organics, etc). I'm having a hard time imagining any of the farmers that I know shelling out a couple of grand on a robot with a camera, to run up & down the fields.

You must not be very in touch with your local farming community. Most farmers are quite willing to try new things. For example, my own father (independent Midwest family farmer, works some 2000 acres with my older brother) has been doing no-till for over a decade. He's also invested heavily in equipment and software for mapping out yields for each field, and more. However, you have to realize that farming is a business. Everything is about the bottom-line. As such, there's no more money to be made in organic farming (less money, actually), so it's not something he does. Leave that to the hippies and the yuppies that don't mind paying $5.00 for a pint of milk.

Then again, if they can show how the cost is offset by gains in yields, then it just might get some use.

This sounds more like a labor-saving device than a yield-increasing device. As such, I doubt you'll be able to show an increase in yield, and certainly not one significant enough to justify this technology. Instead, you should be looking at savings in wages. If the cost (purchase cost plus any ongoing maintenance costs times the expected number of units needed) is less than what it would take to pay minimum wage to some amount of local kids for a few weeks of summer work over the expected lifespan of the robot, then maybe it would be worth doing. On the other hand, it's probably a better idea to just continue hiring kids to walk the fields. Everybody wins -- the kids get money, excercise, and a nice tan, while the farmer gets cheap labor and clear fields, and the community gets something for teenagers to do during the lazy summer days rather than get in trouble. And if you do it right (ie, use hooks to cut out the weeds, rather than herbicide sprays), you won't even damage the environment by introducing herbicides to the food or the ground water. (can you tell I spent most of my summers from around age 8 to age 15 walking fields for my dad?)

Another concern is battery life - just how far is this thing going to go on a charge? 1 mile? That'll cover 4 rows...then what?

If they're serious about this technology, it won't be battery driven. It'll be driven by diesel, and probably will be expanded to cover multiple rows at a time. Also, you can't really convert miles into rows. Depending on the geometry of the field, one mile could be two rows, or it could be twenty.

I do see one potential problem with the technology, though. Since it uses video recognition to determine what's bad and what's not, what happens when a plant is only bad in a certain scenario? For example, corn in a soybean field is considered a weed. This robot should kill it, or the robot will be worthless. However, corn in a corn field is not a weed. If the robot somehow determines that the cornfield is actually a bean field, there goes your entire yield.

Eliminate one problem; another will appear!

[standards]

The problem is that there will be a class of weed that this thing won't be able to "see"... and that weed will thrive.

THEN you'll have to hire a worker to go pluck them out. Or get a software upgrade. Or both.

Good.

[afidel]

Anything which can reduce the reliance on pesticides is a GOOD thing. Now we just need to do something similar for fertizilers, if a farmer could use a lower base level of fertiziler and have the machine add additional amounts only to those areas that most need it then the overall usage would probably go way down and the impact on the environment would be reduced.

weeds aren't the problem, weed killer is.

[aethera]

This gives factory farming a new name!
Why do we get weeds? As they say, Nature abhors a vacuum. So, any tilled space between plantings and rows is enrgy going to waste. Weeds sprout up in this empty space to capture that energy.
So if you want to control weeds in a manner that doesn't cost (hundreds) of thousands, pollute rivers, stream and groundwater, just keep it simple stupid:
Plant cover crops in-between and among your primary crop. It could be a harvestable plant, such as pole beans on corn or basil with tomatoes, though this makes harvesting a job for people and not machines. Or plant a companion crop which adds nutrients to the soil. Legumes add nitrates, buckwheat grass makes great compost, just till it under with your next planting. Even better plant flowers and other hebs that attract colonies of beneficial insects that will help control insect populations in your primary crop.
We got by for a long time without these chemicals. Organic farmers in the US and Bio-dynamic farmers in Europe and harvesting yields that dwarf factory farms, with better flavor and nutrients than conventional produce, and no toxic chemicals.

already been done

[maxpublic]

This idea was implemented ages ago, using a device called a "gardener".

No story here.

organic!

[xeno]

What a waste of technology! Think about it: They're going to go to the effort of doing a brute-force identification of weeds (i.e. looking at every plant in every row, instead of using a broadly-targeted agent). But then they fall back on an old-school method of killing the weed -- by poisoning. Yeah, targeted poisoning, but it's a refinement of an undesirable technology.

Now, it seems to me that if you're going to the effort of imaging the leaves of virtually every stalk coming out of the ground, you ought to take the opportunity to do something very selective and low-impact to the surrounding plants. Like pull the damn thing out, roots and all. No expense of chemical agents, no breeding of resistant weed species, and it produces a product that people are willing to pay a premium for. I'm pretty sure that having weeds pulled robotically would not impact the qualification of a product as "organic." :)

Just my $0.02US, but I'd sooner feed my kids methodically/robotically well-managed organic food than feed them foodstuffs protected by well-targeted herbicides and pesticides. Why play that lottery if you have the technological means to avoid it?