AI being helpful thread

   / AI being helpful thread #141  
You would think it would be simpler to train it what it SHOULD look like and to notice when there is a difference.
Well, a controlled factory is a lot different that a construction site. Trash piles, weather, out of square, imperfect materials, errors in initial set up, ect. Not to say it can't overcome these, but factory automation is easier
 
   / AI being helpful thread #142  
Well, a controlled factory is a lot different that a construction site. Trash piles, weather, out of square, imperfect materials, errors in initial set up, ect. Not to say it can't overcome these, but factory automation is easier
That's true and another way to tell the bricklaying video was just standard automation. Notice that it's only laying bricks in a straight line and there is a solid material path on the ground for the robot to roll back and forth on. It's not dealing with corners, curves, debris on the ground, etc. If it's AI let's see it build a stone wall around a corner with as-found stone shapes on sloped ground.
 
   / AI being helpful thread #143  
If it's AI let's see it build a stone wall around a corner with as-found stone shapes on sloped ground.
As someone who has spent a lifetime living in and repairing old stone houses, I'd guess it's probably easier to build an AI-powered robot to do this work today, than it is to find a competent and available stone mason in the last few years. :p All of the good ones around here are retirement age or older, and all their schedules are fully booked if they're at all reasonable and good.

It was already a dwindling trade 20 years ago, but the housing meltdown of 2003 - 2008 caused nearly all of the experienced guys to have to lay off their younger apprentices, and it seems very few ever recovered or got back into the game. The result is a decade-plus interruption in the creation of new talent, that was already too scarce.

Block masons are easy to find, as well as guys who work in pavers... dime a dozen. And anyone can do stone facade work, or lick and stick crap over stick-frame construction. But guys who actually knwo how to build in structural stone are not easy to find. When you do, they're often booked out a full year.
 
   / AI being helpful thread #144  
This is a good example of why experts are predicting an AI market bubble. Companies like this co-opting the term "AI" in their marketing when in reality it's just standard automation. It's an industrial robot with vision guidance and proximity sensors - pretty standard stuff. The problem is investors are dumping huge sums into these companies because of the misappropriation of the term AI.

Before retiring in 2021 I was an engineer in the factory automation field and did robot and vision system integration. AI is just now starting to be applied to vision systems and it eventually will be a game changer but it's not there yet. Humans are amazing at identifying things, especially defects. Current vision systems can only identify defects that are specifically trained and when you're building a machine for a new product it's hard to get realistic defects for training the system. You end up intentionally creating defects but they're rarely the same as ones that occur naturally. And there are always defect types you didn't consider so they don't get trained into the vision system. AI will eventually tackle this and it will make life a lot easier for the engineers and technicians who maintain machines with vision. Tweaking vision is a frustrating, non-stop job in factories.
First you say it's just standard automation, then you say AI is just starting to be applied to vision systems... which is exactly what this company has done. Plus they're using it in other aspects of the total project, not just the vision system.... it's using AI.
 
   / AI being helpful thread #145  
First you say it's just standard automation, then you say AI is just starting to be applied to vision systems... which is exactly what this company has done. Plus they're using it in other aspects of the total project, not just the vision system.... it's using AI.
Nah. Bricks are uniform in size so they're easy for vision to recognize. The wall is in a straight line so it's easy to follow both mechanically and with vision. The bricks aren't being picked out a random pile - they're being fed to the robot in a very controlled way so they can be picked up repeatably and the robot will know exactly where its edges are. It uses the exact same amount of mortar on every brick which is very easy to automate. It's rolling on a clear path on a mat. Everything about this is easy to do with standard automation. There's no AI happening here except a company using those two letters to get investors.
 
   / AI being helpful thread #146  
Nah. Bricks are uniform in size so they're easy for vision to recognize. The wall is in a straight line so it's easy to follow both mechanically and with vision. The bricks aren't being picked out a random pile - they're being fed to the robot in a very controlled way so they can be picked up repeatably and the robot will know exactly where its edges are. It uses the exact same amount of mortar on every brick which is very easy to automate. It's rolling on a clear path on a mat. Everything about this is easy to do with standard automation. There's no AI happening here except a company using those two letters to get investors.
So now you're saying they aren't using AI in the vision system?
 
   / AI being helpful thread #148  
So now you're saying they aren't using AI in the vision system?
Doubtful because there wouldn't be any need. Locating a simple rectangle like a brick is incredibly easy. Big players in machine vision like Cognex and Keyence are building AI into some of their systems but it's a feature you pay extra for. You wouldn't pay extra for it in this case because it's such an easy application of machine vision. You spend that money on the tough problems.
 
   / AI being helpful thread #150  
Read the article
I did. The article claims they're using AI without a single detail. It doesn't say what problem AI is solving, what decisions it's making, or why AI is necessary for such a straightforward automation problem. Just the simple claim that they're using AI. It looks like yet another company using the term AI to rope in investors. There are real problems being solved by AI. This isn't one of them.
 

Tractor & Equipment Auctions

2008 INTERNATIONAL WORKSTAR 7400 SBA 6X4 DUMP TRK (A52706)
2008 INTERNATIONAL...
Magnolia Trailer with Adams 25 Ton Fertilizer Tender (A55315)
Magnolia Trailer...
White Metal Roof Panels (A55758)
White Metal Roof...
2019 Allmand Night-Lite V-Series 7 kW S/A Towable Light Tower (A55973)
2019 Allmand...
2018 Dodge Charger Sedan (A56859)
2018 Dodge Charger...
2016 TRAILSTAR 40FT DUMP TRAILER (A56129)
2016 TRAILSTAR...
 
Top