[–] individualin1984 0 points 14 points (+14|-0) ago 

I love the idea of this, but they are still stuck on one robot replacing the farmer in the driver's seat. Untill they embrace a new dynamic that robots can bring, this research will only crawl along. The future of automated farming is in many smaller task designed robots not in automated old school farm equipment. They would be much better served grabbing a bunch of farmers and talking through better ways of doing things if today's farm equipment never existed.

[–] rwbj [S] 0 points 6 points (+6|-0) ago  (edited ago)

The article mentioned this, precisely:

While it’s possible that at some point there might be significant labor savings by fully automating farming like this, there are lots of other, more immediate benefits. With fully autonomous farm vehicles, you can use a bunch of smaller ones much more effectively than a few larger ones, which is what the trend has been toward if you need a human sitting in the driver’s seat. This means higher precision, minimal soil compaction, cost savings, and increased flexibility to deal with mechanical breakdowns. Without the need for daylight, you could also keep a farm active 24/7 with a very small human workforce just there (or even checking in remotely) in a supervisory capacity.

Like all things in 'AI', once it's achieved - we take it for granted and shift the goal posts. This project just proved that it's possible to autonomously complete an entire farming season from planting to harvest without ever taking a single foot in the field. Most people would probably agree that's very possible. However, 'very possible' and 'done' are two very different things. This was a moonshot proof of concept that succeeded. Now that it's proven that it can already be done with present gen technology without all that much difficulty comes to the more refined implementations and commercialization.

[–] Naught405 0 points 2 points (+2|-0) ago  (edited ago)

We're a generation out from that, we need people familiar with automation and people familiar with robotic farming. We also need commercial interest and funding. Right now the commercial interest and funding is for putting a robot in the driver's chair, so that's what automation specialists are doing for them. Retrofits are often the first stage.

[–] individualin1984 0 points 1 points (+1|-0) ago 

Yes, but that drives me crazy, when I seems obvious where this should go.

[–] AngryShithead 0 points 1 points (+1|-0) ago 

It's moreso a task for engineers that familiarize themselves with agricultural processes.

[–] PraiseIPU 0 points 1 points (+1|-0) ago 

Its doubtful someone in the industry would have much to bring to the table..

The old gaurd in most fields is what holds progress back..

It takes a new generation and people outside the industry to make significant changes..

Even when the new way means saving lives people will hold to their old ways because they are to damn lazy to learn something new.

[–] prairie 0 points 0 points (+0|-0) ago 

I guess hydroponics can only do so much? That seems like a far more controlled environment for growing food.

[–] Caesarkid1 2 points 10 points (+12|-2) ago 

It's interesting watching people cheer a way of life coming to an end.

[–] Whitworth 2 points 10 points (+12|-2) ago 

Automation of farming doesn't take jobs. Whether a land owning farmer drives a tractor or turns on a robot, he still owns the land. The only ones losing jobs through automation of agriculture are migrant workers.

[–] goat2017 0 points 12 points (+12|-0) ago 

It will end with massive scale monoculture farming run by an agricultural corporation. Having individual landholding farmers becomes inefficient when the whole process is automated anyways.

[–] tame 0 points 1 points (+1|-0) ago 

The only ones losing jobs through automation of agriculture are migrant workers.

Seriously? How many people who work on farms are landowners vs. low-skilled workers, do you think? Hint: Most people who work on farms don't own them.

[–] library_of_stupid 1 points 7 points (+8|-1) ago 

Wait until AIs start taking jobs from journalists, doctors and lawyers, then the screeching will begin.

[–] PhilaFerret 0 points 2 points (+2|-0) ago 

Paid actors once complained about CGI (mid-90s) but it quickly died out, likely their agents told them to can it.

[–] SO_TRIGGERED 0 points 1 points (+1|-0) ago  (edited ago)

Machines can't take jobs from professionals, human judgment will always be needed, and if anything automated data crunching will make them more effective since human memory is fallible, and field knowledge is always advancing and it's easier to teach a machine than a middle-aged person.

Obviously this exempts journalists since they're already not professionals and I look forward to them being unemployed since they're muckrakers and stuffed suits.

[–] B000ne 1 points 0 points (+1|-1) ago 

Your comment had me thinking. Will we have to amend our laws when AI starts using loopholes we couldn't begin to imagine? Or will it just wipe us out

[–] 123_456 2 points 4 points (+6|-2) ago 

Farming isn't fun. It's hard laborious work. Making food cheap, and plentiful, easily, no matter who loses a job, is a good idea.

[–] SorenK1844 4 points 0 points (+4|-4) ago 

"it's hard laborious work" - yes thats a good thing. Any other type of work is a surrogate, and not useful. This robotics shit needs to stop.

[–] twee 0 points 3 points (+3|-0) ago 

Look at it even further down the line. Once our food production and other services get automated, that means that the national economy doesn't need our direct and active input to grow and produce wealth which will then create the possibility to give a "guaranteed minimum salary" to those displaced by the automation.

I honestly think that, if we don't kill each other over bs, we'll end up with the same type of economy that the Star Trek universe is theorized to have.

[–] Crashmarik 0 points 1 points (+1|-0) ago 

Trade in Holodeck templates of Vicotria's secret models ?

[–] HarveyHarveyJones 0 points 3 points (+3|-0) ago 

"Why is there a bolt in my beer?"

[–] AssFaceSandwich2 0 points 5 points (+5|-0) ago 

Would you rather find a finger?

[–] NeoMarxistZoomDweeby 0 points 2 points (+2|-0) ago 

Nah. A lot of this was automated anyway. I knew a lady over ten years ago whose brother still ran the family farm in Iowa. All of his equipment utilized satellites to till the earth, sow the seeds and harvest. The only difference I see here are the "scouts" that take soil samples back to humans to analyze. It still requires human intelligence to program the machines and actually make the decision on when to sow, etc.

What the article isn't saying is how TRULY autonomous AI is nowhere near the horizon. The complex problem of navigating an environment independently has proven far more complicated than they anticipated. People often don't realize that an incredibly complex set of mathematical equations are taking place constantly just to move ourselves from point A to point B. Something as simple as sitting in a chair is incredibly complex, something highly dependent on things like the shape of the chair, the lighting in the room, the height of the chair, etc. and that shifting your perspective just a few degrees completely alters all of those things.

Nobody notices that in all of those robots they see from Boston Dynamics, they are being controlled by remote by a human. They're having a difficult enough time just getting robots to move in a vaguely similar way to humans, much less getting these robots to move fluidly autonomously.

Then you have the issue of consciousness. They have not a single fucking clue as to what the hell it even is, much less how to recreate it. As it stands right now, their hope is that they create it without understanding it.

I think a great deal of this "robot overlord" shit is meant to scare people into staying in mundane, shit jobs with low pay and not complaining or doing anything to demand better pay or conditions. They just hold it over our heads that "we will replace you with robots" and people suddenly put up with shit they wouldn't otherwise.

[–] rwbj [S] 0 points 4 points (+4|-0) ago  (edited ago)

You're mixing some truths with some falsehoods here.

In the most general sense there are two major parts to automation - hardware and software. The bottle neck used to be software, and vision in particular. That's rapidly no longer becoming the case. AI systems very recently managed to overcome even humans in object classification/categorization. And again that's based on pure visual image recognition without any hints or cues.

The new bottleneck is robotics. I can write a program abstracting away all tasks involved in the vast majority of basic jobs. However, getting machines with the agility and reliability of humans is still a problem that's being worked on. Boston Dynamics being remote controlled is not like you think. In our imaginary scenario of sitting down, the person controlling the robot is not telling it exactly what to do with each part of its body along the lines of move this body part this far and in this direction and then this one in that... . Their interaction to the robot is going to boil down to something like 'perform action sit at target chair.' The robot then carries out the action. You are correct that even basic tasks like that are incredibly complex. They involve millions of calculations with dynamic updates on a millisecond resolution. The human remote control is control in the same sense that a conductor controls an orchestra.

It's incorrect to say that problems like navigating environments proved far more complicated than people anticipated. In reality I think it has been the opposite, at least in terms of progress. 10 years ago nobody could, with reasonable confidence, say that this would be happening today. It'd seem somewhere between fantasy and absurdity. Progress has been accelerating vastly faster than most expected. It's this progress that's leading individuals like Elon Musk to become a bit uneasy about the future. It continues to come faster, much faster, than we've anticipated.

Back on the farming issue. I think a general way to sum up things is that things that are very easy for humans tend to be very hard for AI systems. Whereas things that are very easy for AI systems tend to be very hard for humans. Creating the robotic and other systems that were able to farm, from planting to harvest, is an incredible proof of concept for full autonomy. Going the next step would require a good bit more (for instance developing a deep learning system to detect when a crop is ready to be harvested), but generally be much easier than what was already achieved.


I suppose I'd count myself as a part of the ominous "they" you keep mentioning. The reason I'm so excited about robotics is certainly not to hold anybody down, but to free people. When farming can be 100% automated, it would be trivial to create massive farms capable of feeding the entire country (and in the grand scheme of things even the world) without any human involvement whatsoever. Foodstuffs can become effectively, if not literally, free. I think that human labor should be dedicated towards things that humans themselves want to direct their labor towards. If somebody gains fulfillment from farming itself then there would be nothing stopping them from setting up a 'human grown food' farm - there would certainly be a market for it! As we cut the costs of basic survival down to something approaching $0 entrepreneurship becomes ever more possible. Right now you could never setup that farm because you wouldn't be able to afford your basic living expenses, the least of which not being food. As these expenses disappear, infinite opportunities and possibilities emerge.

[–] prairie 0 points 3 points (+3|-0) ago 

During the Hands Free Hectare project, no human set foot on the field between planting and harvest—everything was done by robots.

The thing is that you don't even need to automate 100% of it to get the most benefit from automation; that last 5-10% costs a whole lot to automate (robot maintenance, dealing with special soil issues, etc.).

[–] captainstrange 0 points 1 points (+1|-0) ago  (edited ago)

It still requires human intelligence to program the machines and actually make the decision on when to sow, etc.

Neural Programmers are making major strides in this area.

The complex problem of navigating an environment independently has proven far more complicated than they anticipated.

Systems utilizing random forests, heuristic models, unsupervised learning are rapidly solving this.

People often don't realize that an incredibly complex set of mathematical equations are taking place constantly just to move ourselves from point A to point B

The PMC (premotor cortex) does a lot of the planning for this, and many of the systems I just mentioned are being used even now in industrial research labs to emulate the pmc, particularly in places like Boston. We even have models of what is going on in that regard, running on tf networks, and people like Jeff Hawkins have gone a long way explaining the processes involved, bringing the research together and simplifying what is otherwise a very complex field. In fact, if functional connectivity has any grain of truth to it, it may well be that these 'equations' and 'planning' may be effects that arise from the brains attempt to synchronize different regions of the brain.

Nobody notices that in all of those robots they see from Boston Dynamics, they are being controlled by remote by a human.

I have a very good reason to say you don't know what you're talking about here unless you work in the same lab. Early on, yep, for sure, but the leg planning algos are surprising understandable, and I can say the systems are well past that point. The robots are using a two part system, running in parallel, one for balance, and the other for path planning, sort of like how an octupus has a regular brain and then has nerve bundles or 'mini brains' in each tentacle. What is happening is the legs work together to prevent falling, while the planning system decides the best course and safest paths--and can even override the legs to go across a path that the legs might avoid.

The legs keep the machine balanced using a GAN (generative adversarial network). I won't explain how that works because you're probably already familiar, but the idea is that we invert the networks goals. You see, for complicated problems where the solution space, or 'correct answer' set is huge, such as crashing or falling, the problem can be hard, because the solution space for what we want to happen, not crashing or falling, is much smaller, many orders of magnitude in fact.

What we can do, and what is done, is we train the gan to recognize and predict across the larger solution space, and then have it select for the least optimal answer--learning to WALK by FAILING to fall.

Then you have the issue of consciousness. They have not a single fucking clue as to what the hell it even is, much less how to recreate it. As it stands right now, their hope is that they create it without understanding it.

The science as far as that is concerned, is no longer in infancy--development is at this point middle, or late 'toddler' stage. Theoretically we don't understand a lo of what we are doing, but pragmatically we grasp some of the basics and are working towards the first stage of implementation. For example there is Conant's work on consciousness independent of his regulator paper, a bunch of researchers out of Japan who have successfully created models of self reflection, then theres FCM, and one of my favorites, Conant and Ashby's "every good regulator of a system must be a model of the system it regulates."

Early consciousness in all the basic principles, looking at a wide variety of predictions, and using geometric mean estimates, is just shy of a decade away. This isn't one of those "oh, we'll have it..probably..in ten years! For sure..maybe 20!" type deals, like we have with nuclear fusion, cures for cancer, etc. This is a "definitely 7-10 years, and commercialization at most 1-3 years after that" estimate.

I think a great deal of this "robot overlord" shit is meant to scare people into staying in mundane, shit jobs with low pay and not complaining or doing anything to demand better pay or conditions.

There is an ongoing technological arms race for AI, the equivalent of a new Manhattan project, and I'm really surprised more people aren't aware of this. It's an open secret.

They just hold it over our heads that "we will replace you with robots" and people suddenly put up with shit they wouldn't otherwise.

This is exactly the plan, and not an idle threat.

[–] NeoMarxistZoomDweeby 0 points 0 points (+0|-0) ago 

Perhaps you're right about the mechanical engineering, I don't know a lot about that. But I vehemently disagree with on the consciousness issue.

Nobody knows shit. I've studied this extensively for years, in very well aware of the research and theories, and nobody knows a damned thing about it.

[–] Thrus2 0 points 0 points (+0|-0) ago 

I agree with the fact that the robots are just replacing the drudgery of the work where they sat in the cab overseeing it running itself. As every field is different especially as you spread cross state/country/continent/world, then you have tons of different versions of the same crop that has been modified a bit to be better for here vs there, some locations have strange weather patterns that do not behave like most normal places. Jimbo the farmer grew up farming and either worked this land all his life so knows all of this or has local workers that can advise him on it, the AI would need to have the history of weather in every single location to compete with human farmers. If the weather sites are anything to go off of we still can't accurately predict the weather on a large scale so the odds of an AI doing it are not that great.

In regards to moving like a human I have always wondered why? stick the fucker on tracks and just have it clear the doorway move out of the way and then stop it doesn't need a chair to sit in it would be just as happy to stay upright in fact not sitting down would mean less moving and less wear and tear so is better to stay as is. We are obsessed with making humanoid robots so they can do our tasks but not force us to change the equipment, if we are not going to do the job anymore why not change the equipment to be best for automation instead of best for a human?

[–] tanukihat 2 points 2 points (+4|-2) ago 

And suddenly Mexicans were no longer needed in the world.

[–] library_of_stupid 1 points 4 points (+5|-1) ago 

They weren't needed before either, all they did was benefit the big guy and fuck over Average Joe by lowering pay.

[–] lord_nougat 0 points 0 points (+0|-0) ago 

iAY CARAMBA!

[–] Thrus2 1 points -1 points (+0|-1) ago 

As long as they will do the work cheaper then a tractor why would the farmer replace them? This tech isn't going to make the tractors cheaper or cheaper to maintain so that cost remains or increases with advanced tech strapped on.

[–] captainstrange 0 points 1 points (+1|-0) ago 

The current future of automation as far as 'replacing humans' goes is this: We will fully replace managers before we replace low level labor. Basically the difficulty of the task is counter-intuitive, the things that seem 'difficult', like analysis, management, and so forth, are all easy. The things that seem easy to automate like manual labor, are very difficult.

I mean yeah sure, we'll automate manual labor pretty quickly, but not before automating management. You'll have a bunch of people taking orders from headsets, carrying out the systems plans, while the ai watches from a set if cameras or drones or something--or maybe using rfid transponder bracelets to track where workers are and what tasks have been done.

[–] AssFaceSandwich2 0 points 1 points (+1|-0) ago 

In case you didn't see it down the thread, here it is again. Possible future of farming. https://www.metropolisfarmsusa.com/about

[–] dbvapor 0 points 1 points (+1|-0) ago 

I need me some weed growing bots asap.

[–] lord_nougat 0 points 2 points (+2|-0) ago 

My garden seems to grow weeds just fine all on it's own!

load more comments ▼ (17 remaining)