I’ve had quite a month, again, covering clever software and gadgets that continue to inch their way into performing tasks once reserved for humans. These tasks range from mundane material handling to highly skilled engineering design. It has made me think quite a bit about how our world of manufacturing and engineering will be affected by all this artificial cleverness. It is not just hardware-based robots moving material, assembling parts, and machining parts. Advanced software bots, under the umbrella category of “generative design,” are replacing engineering functions, proposing designs for brackets, airplane wings, and vehicle chassis.
Should we be worried? Will these self-driving machines (and engineering software) replace those of us who make our living in the industrial and manufacturing workforce?
That would be the natural human response. Advances in smart software, robotic automation, vision systems, and artificial intelligence (AI) seem to be accelerating. It is easy to find dire predictions. I write this having just returned from a tradeshow, the ATX West exposition in Anaheim, CA, where I saw literally acres of ever-smarter robotics and intelligent automation. I even saw a robot play table tennis with a human, holding its own against a pretty good player. Collaborative robots, or cobots, are commonplace, often equipped with vision so they can see and respond to flexible circumstances, much like a human (see my feature on page 58.) How can we predict what will happen to us?
I’ve learned that the type of predictions a person makes depends on their personalities, and I tend to be an optimist. In the case of robots and automation, I think an optimistic view is warranted, with some warnings. Optimistically, engineers and manufacturing professionals will remain in high demand. A warning is that their work is going to change. They won’t be replaced but enabled—and in that enabling they will need to reconsider their value in the manufacturing process.
A good example is generative design, a set of software technologies I also cover that help engineers specify a design. One example is topology optimization, a software technique offered by many vendors that mathematically finds the strongest physical design with the least amount of material. The human proposes an initial, blocky shape, specifies the loads and material properties, and the software iterates until it comes up with the lightest solution. Often, the resulting shapes look organic, like something nature would produce. Additive manufacturing can produce these organic shapes, or the shape can be reworked to fit traditional manufacturing techniques. With slight tweaks to outputs and thresholds, the computer can produce hundreds of designs of startling complexity much faster than a human with a CAD program.
This lesson in generative design is important: the human design engineer needs to carefully specify the inputs and evaluate the output. He or she needs to be an expert in choosing those inputs, understanding their practical constraints, and evaluating the outputs to select the best design—one that meets criteria only a human can understand. The upside is it would have been impractical for a human to design some of these organic looking parts, even if they knew how to do it. A clever software bot, working with a human, can produce a better part than a human alone. The bot amplifies rather than replaces the human. This process does require the human to be expert at different, perhaps new things. If they cannot adapt, they become less useful as an engineer.
While at ATX West, I had many robotic vendors profess the sentiment they were not trying to replace humans but allow automation to work in collaboration with them. I think they were sincere. The caveat: humans will need to adapt and change.