QOTD: Feeling Less Than Optimistic About Hands-off Driving?

Steph Willems
by Steph Willems

Sorry for another winter-themed post, but real life has a way of pointing out weaknesses in man’s systems. If you’re enjoying life down in Del Boca Vista, Phase III, this should provide some amusement.

Here goes: It’s a mess up north. Mother Nature’s not asking consent, and the roadways around Casa Steph are dirtier than a roadhouse brothel floor at 3 a.m. The last couple of weeks has brought snow, rain, freezing rain, more snow, and a type of precipitation that can only be described at #4 birdshot fashioned out of ice, hurled straight into your eyes via a bitter wind.

As me about my test vehicles’ driver assistance features, and how these bits of technological wizardry worked in such adverse conditions. News flash: they’re often as good as useless. You know what they say about mammaries on a bull…

(If you haven’t guessed that this is a rant that ends in a question, rather than a simple question, consider yourself forewarned.)

Waymo might have a commercial/semi-pilot project on the go in sunny, warm, and dry Phoenix right now, but I’d love to see how those self-driving Pacificas handle their business up in eastern Ontario. One of the good things about the horrible year of 2018 was that the ridiculous, premature predictions around self-driving/autonomous tech went bust. It’s coming, probably, maybe surely, but not in the truncated timeline touted by its most ardent proponents.

There’s a lot of shit to figure out before Level 4 and Level 5 autonomy reaches streets from coast to coast. The tragic Uber Technologies pedestrian collision last March belatedly opened the eyes of one of the more reckless developers, offering a moment for the industry to pause and take a collective breath. This isn’t Fyre Festival 2.0, it’s dangerous machines on roadways. It’s human lives. These vehicles still need real, live eyes and hands and feet more than we realize, and Waymo is right to go the human monitor route for the time being. Caution and endless testing in all sorts of conditions is required before we cede our roadways to robocars.

Full autonomy in any city of your choice is not six months or a year or two years away, and it’s good that we’ve finally realized that.

Near Casa Steph, in the middle of a sprawling tract of agricultural land, millions of dollars came together last year for the development of a 16 km (10 mile) network of Level 5 autonomous test track, complete with signalized intersections, side treets, and buildings. It’s almost finished, and tech companies like BlackBerry QNX, Ericsson, Nokia, and Juniper Networks lurk behind it. A nearby Ford Motor Company R&D facility created in 2017 has access, too. Soon, a 5.2 km (3.2 mile) high speed oval will join the testing grounds.

The usefulness of this course, besides its dedicated and carefully controlled nature, is the weather. It’s highly variable, to say the least. This city went from tornadoes to snowstorms and an 80F temperature drop in the span of two months last fall — optimum placement to replicate the gamut of conditions faced by vehicles in real-world North American applications.

But that’s the future. All we have now is driver assistance systems that flirt with Level 3 on the high end, and Level 2 in other well equipped vehicles. Radar-guided cruise control, lane holding, automatic emergency braking with pedestrian detection, proximity sensors for parking lot duty, etc.

Radar, cameras, and LIDAR are fantastic things, but fallible — especially if the software governing them isn’t strong willed. We’ve seen that in light rain. When foliage leaves the trees in otherwise dry winter conditions, confusing AVs in carefully geofenced areas mapped out in the summer. How about you add snowbanks of varying heights and shapes to the side of the road? How about if those lane markers are obscured by, you know, snow and ice?

Not just snow and ice, either — on a nighttime trip down Ontario’s Highway 401 (Canada’s most heavily trafficked highway) a couple of weekends ago, I turned on my 2019 Toyota Corolla hatch’s Full-Speed Range Dynamic Radar Cruise Control to give the tech a shakedown run. The weather was cold but dry. The highway surface was clear. Everything worked well — admirably well, in fact, with the car’s electronic brain aggressively taking over the wheel to keep the car centered in its lane as we cruised along at 70 mph. My left hand rested gently on the wheel; just enough pressure to keep the system from freaking out at me.

Do you know what road salt leaves on the road? Yes, white dust and white stains, most prevalent near the edges and center line of the highway. Multiple times, the system shut off unexpectedly after brilliant dust obscured the center line or solid lane marker to a sufficient degree. During a shallow, high-speed curve, this meant you’d suddenly find yourself drifting into the next lane, or off the road, cruise control still activated.

Eventually I had to just turn it off. It held the road well in perfect conditions, but couldn’t be counted on to avoid surprises when things got too dusty.

This week I’m in Subaru’s big Ascent three-row, and the roads couldn’t be worse. Great for putting the crossover’s all-wheel drive and traction control to the test (fun, fun, fun), but not good for its high-tech driver’s aids. Not once this week has lane-holding become available. Thanks to slush and snow, the lines just aren’t clear enough for the car. An amber light in the gauge cluster shows its absence.

Last night, as relatively light freezing rain and ice pellets pelted the Ascent, the car’s EyeSight driver assist system (adaptive cruise, pre-collision braking, lane control) blinked off twice, rendering those aids impotent. The Ascent made sure to alert me that EyeSight was offline, due to too much ice and crap blocking the camera and radar. Hell, even the backup camera was next to useless.

If you’re a Northerner, or even if you hail from the South but hold an interest in autonomous technology, when do you predict we’ll see truly durable, winter-beating AVs enter the market?

[Images: Steph Willems/TTAC]

Steph Willems
Steph Willems

More by Steph Willems

Comments
Join the conversation
3 of 48 comments
  • CobraJet CobraJet on Feb 07, 2019

    Last month I was driving my old Mustang on a two-lane county road. I was approaching the crest of a hill and saw a car in the on-coming lane at the top of the hill. That car was dead stopped in its lane. I went on elevated alert and slowed down, wondering why was this car stopped. As I got close to the crest, a car topped the hill IN MY LANE. Somehow my brain instantly said "that car can't stop and I can't stop in time. Reflexes kicked in and I swerved off the road into a yard, dodged a mailbox and power pole, got on the gas so as not to get stuck in the mushy yard and swung back onto the road, avoiding the accident. The other driver didn't stop. Would a self-driving car be able to figure all that out in a matter of seconds to avoid a mishap?

  • PandaBear PandaBear on Feb 07, 2019

    I'd say 16 years. About the same amount of time it takes to train a baby enough motor and sensory skill to drive. If we keep up with enough R&D and testing we should be able to do the same to AV. The problem would be: can you afford it for every car.

  • Varezhka Of all the countries to complain about WTO rules violation, especially that related to battery business…
  • Carson D At 1:24 AM, the voyage data recorder (VDR) stopped recording the vessel’s system data, but it was able to continue taping audio. At 1:26 AM, the VDR resumed recording vessel system data. Three minutes later, the Dali collided with the bridge. Nothing suspicious at all. Let's go get some booster shots!
  • Darren Mertz Where's the heater control? Where's the Radio control? Where the bloody speedometer?? In a menu I suppose. How safe is that??? Volvo....
  • Lorenzo Are they calling it a K4? That's a mountain in the Himalayas! Stick with names!
  • MaintenanceCosts It's going to have to go downmarket a bit not to step on the Land Cruiser's toes.
Next