Robot hand could harvest blackberries better than humans | Hacker News Reader

Robot hand could harvest blackberries better than humans

news.uark.edu

111 pointsPaulHoule69 comments

Having filled my share of blackberry flats, the first part of the picking motion is a ripeness check. You want the berry to be firm with a bit of give, and you want it to pull free from the plant easily. The force feedback sensors mentioned here seems to be for training purposes, but they would probably be better used on the finished grabbers to detect ripeness.

Yes, this. There's a certain way that ripe blakcberries (and raspberries) feel and pull away from the plant. After a while you just sort of understand the tactile sense of ripeness. So while this might pick blackberries without damaging them, I doubt they're going to be peak ripeness. Blackberries don't continue to ripen after they're picked so it's not like tomatoes where they can pick them kind of green and they ripen up by the time they hit the store.

For blackberries at home or in hedgerows, picking only the black ones has always been good enough for me. For raspberries, it's important that they come off the stem readily.

Maybe commercial pickers need to optimise differently and could pick every day and get every berry at its absolute peak? But I doubt that's the state of the art with human pickers at the moment: they get paid minimum wage and are assessed primarily on the weight they pick.

Yeah but are there other, sensor based approaches that would also work? Ripe berries are softer, so sonic/ultrasonic waves would propagate differently. Is there a wavelength of light which behaves differently? Can shaking the branch and observing the motion of the berry provide a clue?

im pretty sure you can determine ripeness solely from the colour of the fruit for most fruits, no?

There are two metrics - softness and ease of separation. Ideally there would be a 2nd force sensor that measures pull on the berry and then releases the grip if a max value is hit.

That being said, ripeness is best detected by measuring water and sugar content. I'd imagine a more mature version of this tech would be the standard in a decade - https://paw.princeton.edu/article/princeton-engineers-use-wi...

I'm curious if they could combine the force sensors with some kind of micro-movement feedback

Every time I see these headlines, the tech seems to be at least 10 years away from product.

- demos done in a lab controlled environment without the crazy things that happen in a real world.

- no humans nearby so none of the safety features that would be needed should this thing work alongside/near humans.

- no regards for economics, expensive vision models, expensive hardware, no consideration for maintenance and repair costs

This is a classic example of University press releases, you learn to recognize the pattern. Someone who's skill set is PR gets a dumbed-down version of the science, and then converts that into a hype piece that ignores reality in favor of vague statements.

If you want the essence of this technique look at any university press release about fusion technology.

... and if you ask them "got RSS?" they're the least likely to respond.

> no regards for economics

This is the #1 killer every time.

You will always find the most efficient farm machinery to be the least human-like in its design principles. The more it looks like something out of Mad Max the better.

Unless we come up with a machine like the combine harvester for blackberries, no one is going to be interested.

>Unless we come up with a machine like the combine harvester for blackberries, no one is going to be interested.

Good news! https://airharvesters.com/

Clarkson's farm taught me that this is already a thing.

There are several kinds of blueberry picking machines. There are air-blast pickers that blow the berries off the plant. There are ones with wheels of vibrating sticks. There are ones that get a comb around the plant and pull.

Some berries get damaged, yes. Some leaves and twigs get through. They're separated out by a very fast vision-based sorting machine before packing.[1] That's been standard technology for a decade or so.

Apple picking is still in the R&D stage.[2] Cost needs to come down to $0.02 per pick.

It's great to see startups in this area, but the thing has to work. There are too many failed ag robotics startups.[3] Ask "could you pressure-wash this thing"? If there are wires, electronics, and bearings exposed, it's still experimental.

[1] https://www.youtube.com/watch?v=Ica3FLAvPas

[2] https://goodfruit.com/lots-of-bots-video

[3] https://www.futurefarming.com/tech-in-focus/field-robots/cha...

> Ask "could you pressure-wash this thing"?

Is that a necessary requirement? I mean, that would probably damage current harvesters, who are human people.

I mean, there are lots of parts of cars where pressure washing would probably force fluid into the bearings.

Might that not apply to hydraulic, pneumatic or electronic systems too?

(I do get what you're saying though)

Pressure washing a combine harvester.[1] Also trucks, tractors, etc. End of harvest cleanup for a farm in Minnesota.

[1] https://www.youtube.com/watch?v=Q-dOW2wn0rM

Yes, powerwashing would be wanted. That's an IP69K, not too hard to hit with some basic mechanical protections.

Unless you need delicate sensors which need direct contact to samples to work.

Maybe it's not a complete necessity, but generally it's gonna be mixed in with big farm equipment that is power washed. The more you have to "coddle" the equipment the less cost effective it'll be for farmers.

Farm workers generally know how to wash themselves. Still I'd wage good money farm hands have used power washers on each other. Probably work well to clean off work coveralls!

Used to work on a farm, and yes we did do this.

Also, our wash down hoses were basically power washers, but like 10x the volume.

>Farm workers generally know how to wash themselves. Source?

> used power washers on each other

Probably, but this is unsafe - you don't want to deal with high pressure jets of water or flying debris near eyes.

> Is that a necessary requirement?

Strictly it needn't be if it offers an even better solution, but, realistically, what startup trying to introduce a new technology (that isn't cleaning technology) has time to also develop a novel way to clean things? It is such an unlikely scenario that it isn't worth considering.

> You will always find the most efficient farm machinery to be the least human-like in its design principles [...] the combine harvester

Oh? I find my human-based process for separating grain to be of the very same principles as the combine. The specific mechanics aren't exactly the same. For example a combine has a fan, while I have lungs. But the principle — using airflow to aid in separation — is the same.

The sprayer is the only piece of equipment on my farm I can think of that employs a different principle to do the job as compared to how I would do the job by hand.

In most cases if you want to machine harvest you have to design your field around that. A vineyard, for example, that is designed to be machine harvested looks very different from one that is designed to be hand harvested. So if you want to machine harvest blackberries at scale you probably have to plant and manage your blackberry bushes in a specific way to allow for machine harvesting.

> "...plant and manage blackberry bushes..."

Have you ever dealt with blackberry brambles? Because "plant" and "manage" aren't the words that come to mind. "Dominate" and "contain"... maybe?

The quote from the researcher is that one "could [hypothetically] design something that is better than the human hand for that one specific task," which gets turned into "some day this specific device could be better" in the prose, which becomes a suggestion that hey, maybe it already is better! in the headline. Everything published by a Uni PR department is a puff piece, frankly I don't know why they're even allowed here.

I agree, we should not allow PR puff pieces by universities, or by technology companies, either.

Well the economics of farm labor seem to be changing dramatically now, so maybe formerly expensive methods become profitable?

Yeah, this is the classic robotics hype cycle

I am not an "AI will kill all jobs" fan at all.

Although when it comes to precedents, "robots" have (luckily) taken many jobs in agriculture already.

theres something noble about back breaking labour for little pay whilst the owner makes the bulk of the profit.... o wait theres not

>Every time I see these headlines, the tech seems to be at least 10 years away from product.

There's no incentive for the capital class to massively invest in fruit picking robotics when there are tens of millions of exploitable humans on the planet that you can use do the same job for dirt cheap.

The economic balance needs to change for change to happen.

That's why the capital class is overinvesting in AI, because that can potentially replace the higher paid jobs where the labor has leverage and turn them into similarly exploitable workers.

There doesn't appear to be anything unique about this particular soft gripper. This blog post is incredibly speculative and really based on nothing more than the author imagining that a grad student's prototype could some day be a single part in a vastly more complex system. There are entire companies that have spent tens of millions of dollars and man-centuries of work trying to pick only strawberries, and strawberries are a lot more durable than blackberries. Vision, motion planning, and controls are all significantly more difficult than gripper design.

The authors didn't test if the robot hand can harvest better than human. They said it "could one day".

They have not even developed the piece that finds and positions the hand.

>Before the robot can be deployed on farms, the computer vision and positioning technologies that would let it find and reach for berries on the plant still need to be developed.

> ...and farm labor has been limited in recent years.

"Train routes across Germany have been a bit congested recently."

Excuse me but what is new here? Soft grippers have been a thing for decade now, there's evem industrial mass produced ones [1]. There's at least a dozens of labs making similar things each in Sk, Japan and China, how could they even patent this?? (maybe first ever in US?)

As for sensors, there's nothing new there either? There's even cheap open source haptics sensor nowadays [2] by meta, or just companies [3] making them. Collecting data with humans is nothing new either.

They don't even have a product, it's a "could" with the cherry on top:

>Before the robot can be deployed on farms, the computer vision and positioning technologies that would let it find and reach for berries on the plant still need to be developed.

Yes the part that have been the hardest (sensor fusion) haven't even been checked...

Berry picking with robot have been a thing for a while now [4]

[1] OnRobot https://youtu.be/BYAwP7Gn4N4?feature=shared

[2] https://digit.ml/

[3] https://www.fingervision.jp/en

[4] https://youtu.be/zBjNe_s3hys?feature=shared

robots could do a lot of things better than humans... if the robots actually existed... and the problems/bugs/limitations were all worked out... and they had ready access to enough power to do the job... and they were affordable enough for anyone to bother... etc.

It's nice to dream about stuff we could maybe one day have I guess...

Selective plant breeding and robotics pickers are the way forward here. University of Arkansas is the holder of multiple plant patents for better blackberries. New varieties are bred for many, many traits (sweetness, transport, shelf quality, ripening window, etc.)

Prof John Clark likely has invented the berries you've ate: https://news.uark.edu/articles/63163/arkansas-fruit-breeder-...

Harvest costs for fruit are an incredibly important consideration for farms and out of the thousands of potential fruits you could eat, the commercial winners have to be profitable.

There are some awesome opportunities for robotics, computer vision, and ML in agriculture. And if you can reduce harvest costs by 75% like this approach for blueberries, farmers have more market options to select better flavor qualities because the harvest quality goes up: https://extension.oregonstate.edu/impact/innovative-harvest-...

When you automate a system, you end up with at least three systems: the one doing the automated work, a control system, and a supply system to deliver and maintain it.

Automating work, contrary to most discourse around it, doesn't simplify things, doesn't make them more secure, and doesn't reduce employment.

Automating increases complexity non-linearly and makes things overall more fragile, and demand more human labor.

This can only work when population and education levels grow steadily. All these systems we build and collectively call civilization only exist and thrive because they are a reflection of the enormous complexity that self reproducing and complex-language enabled monkeys could engender by burning biomass accumulated for millions of years and using that surplus energy.

When our population numbers, and the complexity of our symbolic systems start to decline, we will bring all of these down systems with us.

Like long haul trucking, only a matter of time until farming is largely autonomous. Companies like Lely, John Deere, and DeLaval pushing far ahead on this stuff.

But can it harvest blackberries cheaper than humans?

And is it going to harvest the free blackberries I'm enjoying when I go on a walk?

I don't think so.

I got schooled about the berries I pick on walks

https://mastodon.social/@UP8/114888883357697194

Always the question I ask myself when I see the videos of pakistanis/indians building stuff with a huge workforce when the same are built in North America with very few people and a lot of automation.

We need a new law that merges baumol's cost disease and wright's law.

Farms don't run on novelty, they run on margins

This doesn't appear to be a product yet.

"Could" doing a lot of lifting here.

It's vaporware structural material.

I would settle for a robot that can kill blackberry bushes. Blackberries cut me so much, every time I go do maintenance on another noxious weed, English ivy which is busy killing all my trees

I’m yet to see ivy kill a tree. I don’t get why people think this happens. It can certainly make them look ugly though.

I’ve seen it a bunch of times, but it’s almost always because the accumulated weight and size of the ivy makes the tree fall over in a good storm

I've seen it and then seen the resulting dead tree fall on my house. Tree was already dead from ivy by the time I bought it. When I finally went back to cut it out, the biggest ivy were like 14cm diameter or bigger.

This is what I mean. Maybe the tree was just dead? How do you know the ivy caused that?

I think that's called a goat

or a pig

I would totally put a goat back there if I was able to. Probably would have to own the goat though because it'd just come back every few months.

But… blackberry bushes are the good kind of invasive!

One day, maybe but not yet. From the article:

>> Before the robot can be deployed on farms, the computer vision and positioning technologies that would let it find and reach for berries on the plant still need to be developed.

Note also that there are no photos of the robot. I think that's because it hasn't even been built yet, let alone deployed.

> The robotic gripper was tested on a range of objects, from hard items like a jar of pears and a can of beans to soft, flexible objects like a bag of potato chips and a T-shirt

This paragraph is weird because the article focuses on blackberries, and none of the example is even remotely close

A soft robot hand using guitar strings as tendons to gently pick blackberries is peak 2025 energy. What stands out is the attention to biomimicry and actual force data from human pickers... that’s not just automation, it’s skill replication.

I'll hold my breath, but this would be fantastic.

We need to automate away the boring/hard jobs.

And then what for all the people who did them?

some other skill free job

They go on to do other things like all the other agricultural labor that was already eliminated.

> Robot hand could harvest blackberries better than humans

Could. But does it ? There are a lot of news with "could" lately. Damn, even the world "could" be a better place.

Farm labour in Arkansas… “ has been limited in recent years. ”

But it still could in principle harvest humans?

As always, _we_ are the product.

cheap labor creates its own demand. once its gone you'll see all kinds of things like this emerging.

It’s a press release for a patent with a lot of “robot arm could…” and “once $THING_THAT_HASNT_HAPPENED happens…”, and the topper:

“Before the robot can be deployed on farms, the computer vision and positioning technologies that would let it find and reach for berries on the plant still need to be developed.”

Keep your long sleeved shirts and overalls handy, because robots are not going to pick your blackberries for you anytime soon.

But it will be cool when they pull it off. I was just pondering the automation of blackberry picking, as they are starting to come on in the PNW, and I tire of getting scratched up.

Most of these ideas look great on paper, are easy to implement mechanically and then look great in the news, but then take years to become practical.

nerds mate, very little real world experience. reminds me of silicon valley chads that come up with apps to solve hunger in africa

Reply on news.ycombinator.com