I was reading the very favorable review of the seventh-generation Chevrolet Corvette in The Wall Street Journal (available here, although it may be behind a paywall) and all its impressive high-tech features. In addition to discussing the outstanding performance and ride, materials selection, construction subtleties, and more, I saw one paragraph that made me really stop, wonder, and then even worry a little. It read:
In order to better calibrate the behavior of the various adaptive driving modes (weather, eco, tour, sport and track) — modulating no less than 12 vehicle systems including the electric steering and magnetic adaptive dampers — the Stingray Z51's 19- and 20-inch wheels (front/rear) are fitted with tiny temperature sensors, because warm tires behave differently than cold tires. But because these sensing thermocouples heat up more slowly than the air inside the tires, their signals go through a special temperature-estimating algorithm before they are processed by the driving-mode head office.
Note: “Z51” is the Corvette's optional performance package.
Whoa… In addition to the now-mandatory tire-pressure-measuring subsystem, this car has an advanced temperature-measurement subsystem in the tires. It's more than just the thermocouples and basic data-reporting, too; it has additional data-processing algorithms, which take into account the different thermal time constants of air versus tire. No doubt this is made possible by the combination of low-cost, integrated sensor analog front ends (AFEs) in tandem with associated microcontrollers and non-volatile memory.
(Side note: I'd be interested in finding out why they chose thermocouples rather than solid-state sensors, as thermocouples don’t endure vibration well, and the modest temperature range here could have been handled by non-thermocouple sensors.)
It certainly is an impressive level of sophistication, and I am sure it provides an extra edge of performance near-perfection. But I wonder: Is all this really needed?
First, it does add some cost, and there's an unavoidable cascading chain of costs that usually crops up, beyond the parts themselves: design-in, qualification, BOM and sourcing, assembly, test, and more.
Secondly, there are post-sale issues. Sensors, as with any component, can fail or drift out of spec. While gross failures are usually detectable for sensors, it's often hard to tell when they are performing within spec or have drifted outside tolerance. So now you need additional algorithms to try to figure out if the reading is valid or not, and more complexity and code. Perhaps the vehicle will be recalled by the vendor because some non-critical, secondary or tertiary-level subsystem needs its code and algorithm adjusted? Will the dealer's test setup have to also be fairly advanced to accommodate testing all these new and unique functions? The list of questions can get pretty long, for sure.
Of course, there are places where more sensors, associated circuits, and firmware can improve safety (airbags and anti-rollover systems are two such cases). But you do have to wonder: Is too much of a good thing perhaps not such a good thing, after all? Just because you can sense everything and try to act on the data, should you? Or is that just setting the stage for an excessive number of headaches in the future?