The state-of-the-art in industrial robots suggests that AI is nothing to be feared.
The Automate 2023 show in Detroit showcased a wide variety of advanced industrial automation solutions, and many of them were branded as “AI enabled.” While there is no external manifestation of artificial intelligence in industrial robotics, the implementation of AI will have a profound effect on American industry. Low-code and no-code robot programming is rapidly becoming the norm, and in the not-too-distant future, we can expect robots that can do multiple tasks on the shop floor and be trained to do those tasks in much the same way that we train human workers.
And, access all episodes of End of the Line on Engineering TV along with all of our other series.
* * *
Episode Transcript:
I’m just back from the Automate 2023 show in Detroit, and this year the show floor was abuzz with one issue: Artificial Intelligence. Everyone was talking about AI, and several companies were promoting “AI enabled” systems.
I hate to use the term “artificial intelligence” both because it’s a widely used misnomer for high-speed computation, and because no one can agree on what the definition of intelligence is in humans, let alone machines.
In industrial automation terms, it has a more distinct and useful meaning. Kawasaki brought a wonderful old Unimate to the show. This is the machine that really launched robotic automation at a GM plant in 1961, and in concept it’s not a lot different from modern equipment. Programmers create an instruction set of specific motions which the robot repeats with high precision, hour after hour. Install a wrench on the end effector, and the robot will tighten that bolt, repeatedly. But it doesn’t know that the bolt is there; it simply inserts the tool into a location in space where the programmer expects the bolt will be, and the machine runs up the fastener.
In robotic terms, AI means a vision-equipped end effector which looks at the scene in front of it, recognizes the bolt, then creates the necessary motion commands to engage with the wrench, then tighten it. It’s a major difference. The AI-enabled machine, in theory, can tighten any similar fastener, on any assembly placed in front of it, just as a human might. This kind of capability results in low-code or no-code robot programming, which is arriving just in time for a resurgence of American manufacturing that is accompanied by a desperate shortage of qualified workers.
Does this mean that the kid pushing a broom can become a robot programmer? Maybe, but it strongly suggests that small and medium-sized operations that buy industrial automation don’t automatically need to hire a degreed engineer to make it work. Robot integration will still be a human function, but I can foresee a role for AI in consulting, where a machine might scan a production process, determine where the best fit for industrial automation is, then specify a robot, end of arm tooling and cost the entire system accurately.
The logical people to offer this would of course be the robot manufacturers. Hire a FANUC or Kawasaki or KUKA to apply their AI system to your problem, and you are very likely to purchase their hardware.
Will AI give us Star Wars type humanoid robots that interact verbally with us? Companies like Boston Dynamics and, notably, Tesla are working on this, but we are still years or decades away from robot chefs, or mechanics, or gardeners. But we are not decades away from universal, general-purpose industrial robots that can do multiple tasks on a factory floor with the kind of hands-on training that we use to prepare human workers for assembly tasks.
And as nice as it would be to have a C-3PO type personal valet, that general purpose industrial assembly robot is much, much more important.
