Autonomous Construction Sites

[John C.]

This came up a while ago but it seems things are getting a little more advanced. This article came through my email feed.

https://www.equipmentworld.com/tech...35a7e056326d5b1b7f&oly_enc_id=9541F2856812F0D

This might not be as far away as people seem to think.

 

[CM1995]

Saw that a article a week or so ago.

Jobsite automation has a very, very long way to go if my iPhone's Siri is any indication. Yesterday I had the phone plugged into my truck using Apple Play so I could use the hands free function for phone calls. Me - "Siri call Joe Dirt", Siri - "There is no Joe Dirt in your contacts list". Now "Joe Dirt" is my superintendent that I had previously conversed with earlier in the day via a couple of texts and a few phone calls.

Automation? Self driving cars? No thanks.

 

[John C.]

I like Mac stuff but all the automation stuff I've seen so far is run on Windows. That's probably scarier.

I still think though that if cars can run down the highway, moving dirt on a construction site isn't going to be that tough.

 

[savman]

That article is nothing more than marketing for SafeAI.

Nearly all of those types of articles are nothing more than native advertising bought and paid for by the company the article promotes. I don't have any industry specific knowledge exactly how publicists work, but the short of it is you pay them boat loads of money a month and they have contacts in different industry publications that they call on to 'suggest' articles that are all but pre-written for the author.

Now I don't know if back in the good old days things were better or not, but I know now that digital media is almost never about revealing the truth about the world (say depicting the state of autonomous construction machinery) and it's almost always about capturing marketing spend and establishing a narrative.

After reading the article I did a quick search on Crunchbase and sure enough SafeAI just raised a Series A of 21 million on June 21. They are flush with cash and what better way to spend it than by creating 'hype' around the company so they can raise a Series B. Plus, they need to incinerate, ahem I mean invest, cash so they can show how much they are growing and need a Series B.

That's how you get years of praise and outright lies in the media promoting a completely fraudulent company like Theranos. Or how an absolute donkey like that Adam Neumann character can raise something crazy like 15 billion dollars. Or how the guy that founded Nikola made a billion dollars with an supposed autonomous electric truck that had a literal extension cord ran to it during the presentation and filmed a commercial with the truck actually rolling down hill because it didn't run.

You could do worse that just assuming everything you read is a lie, but that goes doubly true for anything in the 'tech' sector. And triply true for anything that uses the words deep learning or neural net. If the have AI in the biz name...lol, just forget about it.

Other buzzwords where your BS radar should be screaming: green, carbon neutral, EV, renewable, and anything that mentions climate change.

Yes there are exceptions, but man everything is fake these days. /rant.

Also a skid steer is pretty much the last piece of equipment that would have a justifiable use case for 'autonomous technology'.

 

[CM1995]

I still think though that if cars can run down the highway, moving dirt on a construction site isn't going to be that tough.

Developing a self driving car that can go, stop and turn on a pre-determined route on a hard surface within the parameters of the road. Whole different ball game when when your dealing with something like this -

 

[John C.]

I would thing the autonomous stuff would start with grading a site. Doesn't take much to just load the coordinates, set the load parameters and press the start button. Next might be the excavator digging a ditch. Place the machine where you want it to start, set the coordinates of depth, length and spoil pile and hit start. Program in the end point where the machine will go into park until the next step is down loaded. This stuff is already in use with an operator in the seat. There will probably always be people on the ground because right now I don't see a robot being able to crawl down in the trench and make up pipe, wires or fiber optics. So far nothing has more mobility for that stuff than a human. The point is that the dumb stuff is going to take the humans out of the seat. One only needs to look at autonomous manufacturing to see the future is getting nearer all the time.

 

[savman]

re: autonomous manufacturing. While it'd true strictly speaking 'the future is getting nearer all the time'; it's almost certainly not anywhere close.

The robotic arm was invented in 1959 and by 1961 was in production in a General Motors plant. Ironically the company that invented the arm was called Unimation....an abbreviation for Universal Automation. I'm sure they thought that 'universal automation' was just around the corner when the first arm went into production in the 60's.

Now some 60 years later there have been advances in manufacturing to be sure, but we still have very narrow circumstances where anything approaching autonomous manufacturing occurs. And manufacturing is a controlled environment where almost all parameters stay the same. Very different from earthmoving.

Fanuc has had a lights out manufacturing facility that actually builds robotic arms. It's been in operation since 2001. Now you have to ask yourself...if that's been possible for 20 years, then why isn't it more widespread? The answer is in most cases (and earthmoving has several constraints that lump it firmly into the not feasible basket) it just doesn't make economic sense to automate things.

I don't have time to find it, but there is a really good video of Steve Job in the late 80's early 90's giving a speech while he was at Next computers. He talked about building the machine to build the machine. He was speaking of building autonomous factories in the Bay Area to manufacture Next Computers.

Now if the term, The Machine to Build The Machine, sounds familiar; it's because Elon Musk gave an almost verbatim speech a couple of years ago about taking automotive manufacturing into the next century and building an Alien Dreadnaught factory full of robotics that would revolutionize auto making. He went out and bought the best robotics integrator in the world (a German company whose name escapes me)......fast forward a year and several billion dollars later and he's quoted as saying 'humans are underrated.'

30 years later and the same thing happened. Grand ambitions and lofty goals, and at the end of the day autonomous manufacturing was just vaporware.

If Apple and Tesla can't automate controlled environments; we're all safe for a long time.

 

[AzIron]

I would thing the autonomous stuff would start with grading a site. Doesn't take much to just load the coordinates, set the load parameters and press the start button. Next might be the excavator digging a ditch. Place the machine where you want it to start, set the coordinates of depth, length and spoil pile and hit start. Program in the end point where the machine will go into park until the next step is down loaded. This stuff is already in use with an operator in the seat. There will probably always be people on the ground because right now I don't see a robot being able to crawl down in the trench and make up pipe, wires or fiber optics. So far nothing has more mobility for that stuff than a human. The point is that the dumb stuff is going to take the humans out of the seat. One only needs to look at autonomous manufacturing to see the future is getting nearer all the time.

An architect and an engineer across the hall from eachother cant get plans to match on paper let alone that I have seen a water closet put in the electric room on a set of plans

Autonomous sights have to be planned to a T right now I have not met a gc or project manager that understands dirt work or underground not mention last minute revisions that dont make it in the model till after the hole is dug now try digging a sump in an elevator pit that you cant set an excavator next to and you have a backhoe with the outriggers up reaching full reach with the front bucket full of dirt to hold it down we just got done with a job it was a stake less sight all the cassions were drilled and poured when the steel guys were measuring to fit a crane in something didn't jive the whole building was off 3 feet

Autonomous is a long ways off from being realistically efficient and with unrealistic schedules and stacked onto of eachother trades it's not likely to have an easy start

 

[Kiwi-truckwit]

I think the realistic place for automation to make a meaningful start would be haul trucks on mine/quarry sites. If they can't figure that out, then nothing else will work

 

[DMiller]

Have been seeing articles on haul trucks in mines being fully automated, some excavation equipment as well but not to any extreme levels of quantity, expense at initial is high where upkeep and maintenance to keep the systems functional seems a HUGE drawback. Manhours consumed repairing or updating or even just inspecting and cleaning is almost as high as having operators from what have read. Remote operators are next in line, sitting at a desk in a simulator style multi screen profile where operate the machines miles away.

AI is great until placed in service then it takes a ton of manhours to keep it alive. Human decision making may not always be good but is a far cry from AI that decisions are only based on algorithms.

 

[kshansen]

AI is great until placed in service then it takes a ton of manhours to keep it alive. Human decision making may not always be good but is a far cry from AI that decisions are only based on algorithms.

And just what happens when in the millions of lines of code some one inserts' a ( instead of a [ ? Might not show up until some specific situation comes up.

 

[John C.]

Autonomous haul trucks have been in mines for years. Here are a some articles on that front. I wouldn't say Cat and Komatsu are just a couple of companies making a splash in news media to pull in investment money on a dream that might never happen. I haven't heard that any autonomous mine fleets have been hacked so far.

https://www.mining-technology.com/comment/australia-autonomous-haul-trucks-use/

https://www.caterpillar.com/en/news...duct/a-world-leader-in-autonomous-mining.html

https://www.cnbc.com/2020/07/16/min...utonomous-vehicles-to-improve-efficiency.html

 

[CM1995]

As far as the automation in the auto plants go for every robot building a car there is at least one human looking after each robot on the floor. What robots have done is take the back breaking work out of building a car but it can't do it by itself. We have a large Honda and Mercedes plant in Central AL and robot "techs" are always in high demand.

However building a minivan or SUV is a whole 'nuther ball game than digging around utilities that are seldom marked correctly. Now mines are a perfect environment for autonomous haul trucks but there still has to be humans to program, watch, repair and maintain them.

 

[AzIron]

Cat had autonomous haul trucks at tanaja hills in the 90s but it was to expensive and took a lot of man hours to fix when broke was what I was told at the time

Really on paper i really want autonomous it would theoretically be great but i dont think the Roi is going to be as good as predicted and I bet the overhead for it is ridiculous so in reality the only thing that will realistically bring in autonomous is a labor crunch much greater than what were in witch is a compounded problem cause there is no way to automate pipe installation or concrete placement meaning that operators and equipment will be the least worry