Embedded systems is the most bifurcated of design engineering categories. On the one hand, there are the tried and true -- but decidedly aged -- 8-bit microcontrollers. They're still going strong in everything from servo control to portable medical devices. At the same time, embedded yearns to own the cutting edge; for example, with multicore MCUs in robotics applications.
What's it going to take to continue that upward momentum? That's the question we posed to the Systems & Product Design Engineering group on LinkedIn. Specifically, we asked: What needs to change with embedded systems for real innovation to occur?
William Ketel, an engineer in the Detroit area, set the parameters of our debate by issuing this warning: "Innovation that does not add actual value is a waste of the designer's time."
He alluded to the case of MCUs being replaced by more powerful microprocessors. "There are a whole lot of instances where adding a processor adds features that at best nobody asked," Ketel says.
But if MCUs are the sweet spot for embedded systems, manufacturers often make them needlessly difficult to use. This, in turn, forestalls innovation. That was the take of our own contributing editor Jon Titus. "Many engineers, programmers, and product designers struggle with poorly written, incomplete, and cryptic manuals and data sheets," he says.
The upshot is that hard-to-use tools can raise an insurmountable roadblock to innovation. "If someone has an idea for a product that could use an embedded microcontroller or OEM PC, they face a steep learning curve, which frustrates innovators who might not want to become experts in compilers, CPU architectures, and communication protocols," Titus says.
"The difficulty of understanding how to do what they want with embedded components deters many innovative people from pursuing an idea. Yes, good hardware, software, and documentation exist, but in my experience, not always all from a given vendor. Make the wrong choice and your innovative idea dies."
I concur with Titus. I'd point out that there's certainly innovation in embedded -- witness the emergence of 64-bit and of multicore MCUs. However, we're talking about innovations that apply embedded. In that regard, I'd point out the embedded users are among the smartest engineers and designers around. They have to be, not only for the reasons Titus mentions, but because they're stuck with the tightest constraints around in terms of power budget, cost, and required application performance.
Yet because embedded tends to be an insular niche, innovation within the space isn't apparent to the broader cohort of engineers. As a result, embedded doesn't benefit from the interdisciplinary cross-pollination that's at play most everywhere else.
What's your take? Please leave a comment below.