Vote 2020 graphic
Everything you need to know about and expect during
the most important election of our lifetimes

Why Adaptive Cruise Control Should Be The Last Driving Aid Before Full Autonomy

Illustration for article titled Why Adaptive Cruise Control Should Be The Last Driving Aid Before Full Autonomy
Photo: Raphael Orlove

Everyone these days, from Tesla to Cadillac to Nissan to BMW to Infiniti, has some kind of semiautonomous system that controls the throttle and steers for you on highways. Most of them are exercises in sheer terror. But even if they were good—especially if they were good—they shouldn’t exist anyway.

Advertisement

Let me tell you a brief, boring story. On Sunday I was out on Long Island driving a 2018 Volvo S90, which features Pilot Assist, a system capable of maintaining speed on the highway, slowing for the cars ahead of you, and also steering—on paper, at least. How was it in practice? Terrible!

The S90 bounced constantly between the lane markings, failed to negotiate some curves, and, frankly, made me very scared. The car drove drunk, and a cop would have more than enough reason to suspect the driver was as well.

Advertisement

(When asked, a spokesman for Volvo told me that Pilot Assist was “designed to relieve the burden of the more mundane types of driving such as stop and go rush hour traffic and long interstate drives. It is not intended for speeds above 70-80 mph or curvy roads that should be enjoyed by a fully engaged drivers. You should not have experienced ping ponging if lane markings were bright and clear, and especially not if there was also a car in front of you.”)

To be fair, the lane markings on Long Island, like most New York roads, could be improved, and the road I was on, the Northern State Parkway, is curvier than interstates. That said, Pilot Assist did always keep a safe distance between me and the car in front of me. Also: I didn’t test Pilot Assist in stop-and-go traffic though I’m told it’s very good.

But the shakiness of Pilot Assist on the highway was, in the end, a good thing, as it forced me to be as vigilant as I would’ve been had I actually been driving the car—even more so, but in different ways.

Illustration for article titled Why Adaptive Cruise Control Should Be The Last Driving Aid Before Full Autonomy
Photo: Raphael Orlove
Advertisement

This is true of other systems as well—when I used Cadillac’s Super Cruise system on a CT6 a week ago, it did much the same thing, constantly riding up on top of lane markers before correcting itself and veering back towards the center of the lane. Our experience as a staff has found BMW’s Driving Assistance Plus to be pretty wanting as well. The best of these is Tesla’s Autopilot, which keeps the ping-ponging to a minimum, and actually seems to have a clear idea about where it wants to go.

Advertisement

Therein, though, lies the problem. We can expect that that Cadillac’s system will keep getting better, and so will BMW’s, and so will Volvo, and so will Tesla’s, but with each new update, these semiautonomous systems will become more and more comfortable, and drivers will be left feeling the opposite of what I felt on Sunday. They will feel ease, and eventually, great trust, like the way no one questions the use of basic cruise control. There will be many warnings from manufacturers about how drivers must still remain vigilant, but these warnings will go unheeded, because drivers tend to do what they want.

Illustration for article titled Why Adaptive Cruise Control Should Be The Last Driving Aid Before Full Autonomy
Photo: Raphael Orlove
Advertisement

The better these semi-autonomous systems get, in other words, the higher number of drivers who will be lulled into a false sense of security. Because if we’re not at Level 5 autonomy, there is no real security, and drivers need to still pay attention to the road, even if many won’t.

Which is why I’m proposing that, for real cars that actual humans drive, we stop all semi-autonomous systems at the one feature that does work well—adaptive cruise control, which is a real advancement on a piece of technology that drivers have been safely using for decades. That’s fine until the day comes when cars can fully and completely drive themselves without a steering wheel or pedals. The in-between phase does no one any good.

Advertisement

We should continue development of self-driving cars, of course, but only fully-autonomous should be allowed on the streets. Semi-autonomous cars, as good as they can be, are inherently flawed, a half-assed solution to much larger and more complex problems. It’s a bridge technology anyway, a set of tools we won’t need in 100 years, when, maybe, we’ll look back on in horror at having tried to do it at all.

But I like driving, so I’m glad I’ll be dead by then.

News Editor at Jalopnik. 2008 Honda Fit Sport.

Share This Story

Get our newsletter

DISCUSSION

damnthisburnershitsux
damnthisburnershitsux

My 2017 ridgeline has adaptive radar cruise control and lane keep assist, it will steer for you over 45 mph assuming it can see and lock onto the lines. It does a pretty decent job, however it deactivates after 20-30 seconds if there is no torque being applied to the wheel by your hand.

Take away the automatic deactivation system and this is what is being sold as autonomous.

When I am driving my ridgeline I find the combination of ARCC and LKAS to be really useful on a long 2 lane state route where I might be behind someone for an extended period of time. I don’t expect it to take over completely but it can be relaxing not to have to be burdened to match speed with the vehicle ahead of me and having to apply less pressure on the wheel.

That said I never would want to use this system as a completely hands off system. It just isn’t there. The camera is easily fooled or just doesn’t see the lanes when the road is old or it’s raining or has snowed recently or you are driving directly into the sun or you are driving along a tree lined road in the morning or afternoon when the filtered sun light and shadows plays havoc on the contrast detection system trying to find the lanes. Sometimes if I’m on the highway in the right lane the LKAS will attempt to follow the white line onto an exit ramp as it peels off from the white line marking the edge of the road (sort of like what happened to the model X driver recently killed when his car drove into a barrier at a lane split/merge).

Honda is doing it right by labeling this as an assist feature and shutting it off without constant user feedback.

I should also point out that there is a lane departure warning system LDWS that shakes the wheel and beeps if I am leaving the lane. This system warns of lane departures in situations where the camera is not able to lock onto the road for LKAS, leading me to believe the threshold Honda has placed on the LKAS lock is much higher than the threshold on the LDWS.

Point being there is a lot of judgment on the manufacturer to be applied to the data being gathered and whether they call it autopilot or just an assist feature. I won’t be happy calling it autopilot unless we’re using LIDAR/RADAR or some other scanning feature that goes beyond simple contrast detection camera schemes.