I wrote an in-depth article on what it takes to successfully deploy industrial robots in manufacturing. Hint: robot programming is only a small part of the process. https://www.mecademic.com/en/w…ustrial-robot-programmers
Will AI replace industrial robot programmers?
- Meca
- Thread is Unresolved
-
-
Nation
Approved the thread. -
To replace Robotprogrammers with AI, It kind of requires the customer to accurately describe what they want.
We're safe. HEH
-
Simulation was supposed to replace robot programmers, at least to a large degree: Generate "perfect" simulation, drop into robots, maybe allocate 1hr of "touchup" time per robot, done! It was goig to put robot programmers out of work in job lots.
Instead, the best sim-jockeys are people with hands-on, hard-won robot programming experience on the shop floor, and now the demands on the automation are bigger, more complex, so simulation becomes a necessity just to get a process design to the point where the robot programmers can beat it into working in the real world.
Basically, every time we improve the tools, the demands for what those tools can do also increase. This kind of AI will probably end up a tool in the Simulation toolbox, allowing for path optimization faster and better than currently possible... at which point, every Sim Jockey will be expected to do sims that are 5x as large, 5x as complicated, in 1/2 the time....
-
In my opinion, using simulation tools increase the time it takes to program a robot. Of course it depends on what you are doing but in my experience I am much faster without them. The simulation tools aren't really good enough to easily handle complex systems.
If you are doing complex 3d path generation then that would be a different story.
-
To some degree, I think it depends on the industry. For example, in most aerospace work I've done, it's a legal requirement that every point that touches the airframe be programmed by simulation, directly from CAD. No pendant touch-ups allowed. And when you're trying to drill and rivet some 18000 different positions on a single section of airliner fuselage, doing it by hand would be prohibitively expensive anyway.
But even in "classic" robotic industries, like automotive, I'm seeing a trend to make more jam-packed, complex automation lines, and use simulation and "digital twin" tools to do more in the "virtual" phase, before cells are even released for hardware purchase.
-
Hi,
I have gone through your article and it's really insightful. Thank you for the good read.
Will AI be able to write code is a hot topic right now.
Software developers are more worried about losing their jobs to AI algorithms. No, I don't believe so.
Algorithms aren't as intelligent as human brains. In an Evans Data Corp poll, 550 software professionals were asked about the most concerning parts of their jobs. “Artificial intelligence has replaced me and my development efforts,” claimed 29%.
Though it may sound buzzing. I truly believe these are a lesser concern.
I know it doesn't make much sense. But, that's the truth. To roll out an efficient software you need to test it. The majority of the industry experts claim that there is no other efficient method other than manual testing, to test a software/application. Numerous AI services providers, as a panel, after serious discussion among themselves, have shortlisted the best AI applications that can generate codes. They are
Sketch2Code
DeepCode
PyCharm
Bayou
mabl
Pylint
Clever-Commit
Kite
Embold
Run.ai
The important thing to remember is that these platforms can only generate code, but can never test the efficiency. And these programmers still need human supervision at every stage. So I strongly believe AI can never replace neither the industrial program robots nor humans.
Create an account or sign in to comment
You need to be a member in order to leave a comment