I have this vision of closed-loop feedback systems becoming more important as we demand greater efficiency.
There’s some debate about whether the most important invention of the industrial resolution wasn’t actually the governor on steam engines. A governor keeps the engine running at a constant speed of rotation by adjusting the throttle when the speed is incorrect. Without it, drive shafts would be driven very erratically.
The governor is a simple case of mechanical negative feedback. It mechanically computes the error in rotational speed and converts this into a change in amount of fuel used.
The same principle of sensors providing feedback transformed our dirty, gas-guzzling carbureted gasoline engines into (relatively) clean, efficient fuel-injected ones. It also gives us dryers that can sense when clothes are dry instead of running on a timer. Thermostats form simple feedback loops.
Still, there are a lot of opportunities where feedback could save electricity, resources, or time. Cooking is a great example. You can buy thermometers to embed in things as you roast them, so they come out exactly done. You can buy regulated water baths for sous-vide cooking. Expect ovens in the future measure surface temperatures with IR thermometers and turn off before your food burns.
My dishwasher requires me to experimentally determine the correct amount of detergent for the water here. It could easily sense the mineral content of the water and dispense the right amount of detergent for me. In fact, it may have to in the future to reduce the amount of unused detergent sent down the drain.
The key to all of this is sensors. They’re expensive and finicky. Exotic materials seem like the rule. The oxygen sensor in your car, for example, is plated with platinum (currently over $1000/oz).
However, all we really need is for the sensor output to be well correlated with the thing we want to measure. The worse of a signal we can tolerate, the cheaper the sensor can be.
Electronic sensors are ideal for two reasons. First, we can do a lot of signal processing to extract the information we want from the output signal. Second, electronic systems have virtually unlimited gain.
Back to the governor example, the signal processing corresponds to the way the mass of the spinning balls prevents quick changes in the throttle. The governor cannot respond very quickly, so it ignores brief errors. The gain corresponds to the leverage of the linkage to the throttle: with high gain, a tiny speed mismatch will cause the throttle to open or close a lot. High gain can lead to faster response and better efficiency (though in poor designs it can cause instability or oscillation).
All of this is pretty exciting to a sensors guy like me. If we can just get sensors cheap enough, they will be everywhere. Imagine running with insoles that selectively firm up to correct your posture. Or an oven that knows when your cookies are done. Or a bathroom fan that stays on until the humidity has dropped to normal so you don’t get mildew. There are also sustainability applications, like using soil sensors in fields to make sure we don’t overfertilize or overwater crops.
The applications are endless. Wherever there is inefficiency, feedback can come to the rescue.