How Virtual Should Cars Be?

Driving a simulated car – as in a game – is becoming more and more like driving an actual car – in reality.

At least in terms of the inputs.

The car in the game is steered remotely, via a gamepad. You accelerate and brake the car the same way. Soon real cars will be accelerated and braked the same way. Many already are, at least as far as acceleration. They have drive-by-wire throttle control. Your foot does not actually control the acceleration of the vehicle. A computer controls the acceleration of the vehicle. It assesses data it receives from sensors that are connected to the accelerator pedal, but there is no physical connection between your foot and the throttle.

The good news is the throttle cable can’t stick with drive-by-wire. There is no throttle cable. The bad news is that the computer can “stick” – accelerating the car even though you haven’t pushed down on the accelerator pedal. Even if you’ve taken your foot off the accelerator pedal entirely.

The car industry denies this happens but there is more than a little evidence that it does or at least, has – and regardless, the fact is it can.

The problem is – what to do about it?

Which brings up related questions about other forms of drive-by-wire, including drive-by-wire brakes and steering, both of which are more than speculative problems. They are both in the works, part of the technical development of automated cars – which (ultimately) won’t have steering wheels or brake pedals – or pedals at all, for that matter.

But they are also being contemplated for non-automated cars; I.e., the cars we control. Except, we don’t, really. See the above in re drive-by-wire throttle control. We ask the car – the computer which really controls the car – to perform in a certain way. To accelerate full-on when we floor the accelerator pedal, for instance. But the computer is the Decider.

It can decide to do something . . . else.

Just as in principle (and technical fact) it could decide to do something else with regard to braking and steering the car. And if that happens, there is literally nothing you – the driver – can do to countermand it since there is no physical interface between you and the brakes and the steering wheel.

Two scenarios present themselves, both extremely plausible.

The first is simple wear and tear on the bits and pieces – and the connections. And the sensors and motors that control the operation of these systems. They will be exposed to dirt and moisture, to high heat and extreme cold. They will experience vibration and shock – from hitting potholes and such. There will inevitably be defects or at least, design weaknesses that do not manifest right away – but will manifest eventually. And there will be hiccups and failures related to wear and tear.

All of the above happens to current physical connection systems, too. But total loss of control is extremely rare. You might get a shimmy or vibration through the steering wheel when suspension/steering parts are worn; the car might pull to one side if it is out of alignment. If you lose the power steering pump (or the belt which drives it) steering effort will increase, but you will not lose the ability to control the car’s direction.

What happens when – in a drive-by-wire steering system – the car jerks hard to the left and your attempt to correct via countersteering has no effect – because the computer refuses to recognize you inputs? What happens when there is no steering wheel at all?

With current physically-connected brakes, you might lose partial pressure (the systems of all cars built since the mid-1960s have separate front/rear hydraulic systems precisely to eliminate the possibility of a complete failure) but mashing down the pedal will cause the car to slow down and stop, eventually.

But what happens when the computer doesn’t want to apply the brakes?

Or, when someone else doesn’t want it to?

This is the second scenario – a hacker accessing the car’s computer and causing it to accelerate when you want to stop. Or steering it off the road. This is all within the realm of the technically possible.

Which is pretty damned scary.

Particularly given the fact – as this columnist has observed before  – there is no mandate for a fail-safe, an electronic-age version of the Dead Man’s Pedal which locomotives used to have. Some way for the driver to seize control – if it becomes necessary – from the computer.

This is a safety issue, certainly. One with potential consequences at least as serious as backing up over a kid or having an under-inflated tire blow out at high speed. And yet, no call from the Usual Quarters for “action.”

One wonders, why?

. . .