I don't think we're remotely close to cars being SoCs, or that there's anything about analogue components that magically resists monitoring with sensors.
We're not close, but like I said, we're headed that way. Maybe in another 20 years, say.
The analogue components of a car can, and indeed already do, have out-of-band digital monitoring. That secondary digital supervision feeds both your dashboard, and the pin-outs repair shops connect to. It even feeds things like fuel-efficiency microcontrollers that, in their roundabout way, send commands back to the motor.
But the architecture of even current electric cars goes far beyond this: electric cars don't monitor the analogue components with digital sensors, they simply have ADCs connected right to the sensors, and take the now-digital signal from the sensor and feed it to the car's CPU, where that signal—that data, now—can produce many separate outputs.
Surprisingly (given the last 30 years of robotics), the key advantage of electric cars over petrol ones is that the petrol architecture is inherently subsumptive—digital "cognitive" actions overriding lower-level, directly-connected analogue "reflex" actions—while the electric architecture can centralize data-flow. You don't have to worry about the driver attempting to move the car with the parking brake enabled; the information of "gas pedal depressed" goes through the "brain" of the device, where it also knows that the "parking brake engaged" line is high, so its desired-torque output should just stay at zero.
> You don't have to worry about the driver attempting to move the car with the parking brake enabled; the information of "gas pedal depressed" goes through the "brain" of the device, where it also knows that the "parking brake engaged" line is high, so its desired-torque output should just stay at zero.
You can do all of this centralization in an ICE car. Electronic throttles and switches aren't new. (Though the design I've seen for this case is "oh, driver pushed the pedal, let's disengage the parking brake since they probably just forgot about it.")
> We're not close, but like I said, we're headed that way. Maybe in another 20 years, say.
I don't think we're even headed that way. Not faster than petrol is going to become uneconomical due to renewables or at least government incentives/tariffs.
> But the architecture of even current electric cars goes far beyond this: electric cars don't monitor the analogue components with digital sensors, they simply have ADCs connected right to the sensors, and take the now-digital signal from the sensor and feed it to the car's CPU, where that signal—that data, now—can produce many separate outputs.
This is a statement without meaning. Nothing about petrol engines prevents ADCs connected to sensors.
> Surprisingly ... while the electric architecture can centralize data-flow.
This also seems kind of meaningless.
> You don't have to worry about the driver attempting to move the car with the parking brake enabled; the information of "gas pedal depressed" goes through the "brain" of the device, where it also knows that the "parking brake engaged" line is high, so its desired-torque output should just stay at zero.
This is just drive-by-wire. Again, nothing about drive-by-wire is exclusive to electronic-drive vehicles. For example, all aircraft are powered by kerosene and yet modern aircraft are all fly-by-wire.
There's basically very little about the drivetrain that actually matters to any of this technology. The key advantaged of an electronic drive train that I would highlight, in roughly most to least important, are:
* Reduced maintenance and mess
* Emissions, and access to renewable energy as it increasingly powers the grid
* Regenerative braking (but note that even some analogue vehicles have this in
the form of flywheel energy storage), and
* High torque
But this paragraph of the original article is bunk.
