Oct 31, 2018

Axios Navigate

Good morning and Happy Halloween! Thanks for reading Axios Autonomous Vehicles. Starting today this newsletter will be in your inbox on Wednesday and Friday mornings, delivering news about AVs and analysis from our network of expert contributors.

Please share this newsletter and tell your friends they can subscribe here. If you have tips or feedback just reply to this email. And please take our end-of-year subscriber survey. It's very short and can be found here.

1 big thing: Creepy car behavior

Illustration: Rebecca Zisser/Axios

I’ve never seen a ghost, but it sure feels like I’ve encountered them a lot lately. The other night, an invisible set of hands kept vying for control of the 2019 Acura RDX I was driving. A little tug to the left, a nudge to the right. It gave me the creeps, like the car was possessed.

Cars do scary things sometimes when operating in driver-assist mode: Braking at shadows, swerving away from unseen obstacles, and failing to respond the way we expect.

Why it matters: Consumers need to be able to trust the advanced driver-assist systems in today’s vehicles or they will never fully embrace self-driving cars.

“These systems have a lot of potential for improving safety and reducing crashes. But we lose all of that if they are implemented in such a way that drivers are annoyed or uneasy about using them."
— Russ Rader, Insurance Institute for Highway Safety (IIHS)

The big picture: Safety features like blind-spot detection or backup cameras are the early building blocks of automated driving. We're now seeing more advanced systems —lane-keeping assist, adaptive cruise control and automatic emergency braking — that aim to help drivers or even correct their actions if necessary.

But these technologies don't always behave the way humans would, and sometimes, as I've learned, that can be downright frightening.

  • Cadillac's Super Cruise, a truly impressive hands-off highway driving system, got confused and lurched to the left and back again when it couldn't find the lane markings as I began to drive across a bridge.
  • When I was driving Volkswagen's 2019 Jetta it veered toward the center median when it detected what it thought was an obstacle. It was just a puddle of orange paint spilled by a road crew.
  • IIHS says a Tesla Model 3 often slowed down unexpectedly when it encountered tree shadows on the road, oncoming vehicles in another lane or crossing traffic far ahead.
  • IIHS tested the adaptive cruise control and active lane-keeping assist systems in the Mercedes E-class, BMW 5-series, Volvo S90 and Tesla Models 3 and S. All demonstrated their own share of creepy behavior.

The risk: With names like Autopilot or Pilot Assist, many of these technologies erroneously leave consumers thinking their cars can drive themselves.

  • If companies make these systems too capable, consumers might zone out and not be ready to react when they need to.
  • But if cars brake or swerve erratically, drivers might switch off the technology altogether, missing out on their potential safety benefits.

The bottom line: The early results underscore the fact that today's systems aren't a good substitute for human drivers.

2. Air traffic control for AVs

Illustration: Sarah Grillo/Axios

For the most part, the hype surrounding AVs has focused on the cars: how safe they are, when they'll arrive, whether they'll work. But less attention has been paid to how these vehicles will work together as fleets — in models similar to ride-hailing services, Bestmile CEO Raphael Gindrat writes for Axios Expert Voices.

Why it matters: If AV makers flood cities with driverless vehicles, they could add to the traffic pressures created by badly managed ride-hailing fleets. Efficient deployment will require vehicles, operators and travelers to communicate in real time to match supply and demand.

Background: A recent study by the San Francisco Transportation Authority found ride-hailing services are responsible for 51% of the city’s traffic slowdown over the past six years.

What's needed: To curb crowding, cities could implement centralized control of AVs — like air traffic control for planes — to direct vehicles from multiple manufacturers and service providers.

  • To get all parties onboard, including carmakers and mobility providers, the control layer would need to be a vehicle-agnostic platform that could coordinate and optimize fleets in real time.

Where it stands: The industry is starting to pay attention.

  • A Deloitte article earlier this year argued that cities likely need a comprehensive, interoperable system.

The bottom line: Shared AVs promise to make transit cheaper, safer and faster, with fewer vehicles moving more people. A centralized infrastructure would help realize this promise.

Go deeper: Read Gindrat's entire post.

3. Driving the conversation

I want to hear — and share — what you're reading about AVs. Send me a link to an article and your expert analysis of why it matters — joann.muller@axios.com.

Flying taxi: A pilotless, electric aircraft service plans to start testing in Singapore next year. (Nick Lavars — New Atlas)

  • The big picture, from MIT's Carlo Ratti, who shared the story with us: "The problems of mass transportation can be fixed with our feet planted firmly on the ground — and long before flying taxis are even a viable alternative. And staying on the ground will obviate the need for networks of new infrastructure, like costly 'vertiports.'"

License to drive: Waymo to test truly driverless cars in California (Michael Liedtke — AP)

  • Why it matters: Waymo is the first among dozens of companies testing self-driving cars in California to persuade state regulators its technology is safe enough to permit them on the roads without a safety driver in them.

Point/Counterpoint: Marc Scribner of the Competitive Enterprise Institute responds to former NHTSA Acting Administrator David Friedman's recent Axios Expert Voices post.


Illustration: Rebecca Zisser/Axios

Last weekend, we looked at the myths and realities surrounding autonomous vehicle technology, how driving has changed and what will happen in the years to come.

Afterward, I hopped on Twitter for a Q&A with readers. A few highlights:

Q: How soon can I buy one?

A: The first uses will be in shared fleets in geofenced areas and under good conditions. Buying a personal driverless car is years away.

Q: How will an autonomous vehicle know to obey the hand signals of a flag man at a construction detour? How will it locate and wedge into a parking spot at Walmart the Saturday before Christmas? How will it know where to pull off in case of an emergency?

A: This is why there is still work to do. 99% of self-driving cars is solved. It’s the 1% left that’s so hard.

Go deeper: Read our deep dive on autonomous vehicles and more questions from my Twitter chat.

5. 1 fun thing: Mailbox with a mind of its own

DHL's PostBOT. Photo: Deutsche Post

Several cities in Europe are experimenting with the use of autonomous vehicles for postal deliveries.

The strolling mailboxes provide cost savings to offset declining postal revenue and help lighten the load for human mail carriers.

What's new:

  • Norway's Posten-Norge partnered with Buddy Mobility, a startup based in Oslo and Silicon Valley, to deliver mail and packages.
  • In Germany, DHL's Deutche Post is testing deliveries using an electric delivery robot called PostBOT.
  • The mailbot's sensors track the legs of the mail carrier, following them on their route and navigating around obstacles as needed.
  • At just under 5 feet tall, the PostBOT holds up to 330 pounds of mail and can roll along at nearly 4 mph.

What's next: DHL is also working on a delivery drone called the Parcelcopter.