Nov 19, 2018

Axios Navigate

Joann Muller

Good morning! Thanks for reading. Please share this newsletter and tell your friends they can subscribe here. If you have tips or feedback just reply to this email.

Due to Thanksgiving, we are publishing Monday and Wednesday this week.

1 big thing: Safety drivers are pioneers in AV no man's land

Homer Simpson gets a job testing self-driving cars. Screenshot: Fox's The Simpsons, season 30 episode 5, "Baby You Can Drive My Car"

The job of autonomous vehicle safety driver seems pretty easy: Get paid for sitting there while the car does all the work. But it's a challenging assignment and self-regulated by the companies testing AVs, so the rules are only beginning to emerge.

The big picture: Safety drivers are researchers' eyes and ears, chronicling every roadway encounter to make the technology better. But requiring drivers — even specially trained ones — to pay attention without actually driving is difficult, which is why many companies argue that full autonomy is the safest way to go.

Background: Last March, a pedestrian was struck and killed by an Uber-operated self-driving car in Tempe, Arizona. The human monitoring the vehicle was believed to be watching a television show on her phone.

  • The accident prompted a review of procedures and expectations for human safety drivers.
  • Without federal regulations for AVs — just general guidance — the U.S. Department of Transportation suggests states should be the ones to regulate safety drivers.
  • For now, best practices are emerging from a handful of voluntary safety self-assessments submitted by companies to the DOT.

How it works: Safety drivers typically work in pairs, one in the driver's seat monitoring the environment and the other riding shotgun with a laptop, monitoring the car's computing system and annotating the drive.

  • The driver needs to be capable of taking over immediate manual control in the event of a failure or emergency.
  • Typically, though, human interventions are triggered by a situation that the trained safety driver deems tricky.
  • Argo AI, which is developing self-driving technology for Ford, adopts a "no heroes" policy when it comes to drivers retaking control, says president Peter Rander.
"They need to learn that while we want the car to experience life, it has to do it responsibly. If they’re giving it too much margin and allowing it to get too close, bad things will happen."
— Peter Rander, Argo AI

The work can be physically and mentally demanding. The pilot has to be alert, with their fingers lightly cupped around the steering wheel and their foot hovering over the pedals.

  • AV companies usually limit time in the driver's seat to about 2 hours, and the pilot and co-pilot swap jobs frequently.

AV safety drivers don't need a special permit, but states like California require that they have a valid driver's license and undergo AV training.

  • The training typically lasts at least a month, and involves both classroom instruction and driving on a test track.
  • It includes software and hardware training, including how to turn the self-driving system on and off.
  • On the track, instructors deliberately inject faults into the system to train drivers how to react properly.
"Someone with a computer in the car can cause the steering wheel to make a hard left and that's when the craziness happens."
— Peter Rander

What's next: California and Arizona have already given Waymo permission to test vehicles in their states without a safety driver, and the Google self-driving car unit plans to launch its fully driverless service in Phoenix next month.

2. AV signals are a guessing game

Waymo self-driving vehicles in Mountain View, Calif. Photo: Justin Sullivan/Getty Images

Waymo recently became the first AV company cleared to test vehicles without human safety drivers in California — which has far more congestion than Arizona, the only other state that has granted permits to driverless vehicles, Sudha Jamthe, CEO of IoTDisruptions, writes for Axios Expert Voices.

Why it matters: As AVs make it onto the road, they will interact with human drivers, pedestrians, cyclists and other driverless cars, but there is no industry standard for how they should signal their driving intentions — or even an agreement that they'll have to.

Where it stands: Automakers are experimenting with different ways that cars can communicate intention to drivers and pedestrians.

  • Ford’s AVs have a screen on top of the windshield that blinks in various ways to signal their movements.
  • Drive.ai is testing Nissan NV200 robotaxis in Dallas with LED screens on their sides that display text in English.
  • Uber is exploring AV intention systems that could communicate to pedestrians using flashing lights, sounds and potentially projections.

Yes, but: Waymo, it's worth noting, has not equipped its cars with any special signals to communicate vehicle intent. It has instead focused on preparing its AI for the unpredictability of interactions with pedestrians.

What we're watching: Ford is calling for AV companies, the International Organization for Standardization, and the Society of Automotive Engineers International to collaborate on standards for AV communication.

Go deeper: Read the full post.

3. AVs' uneven effect on car ownership

Traffic on the Brooklyn Bridge in New York City. Photo: Drew Angerer/Getty Images

Axios Expert Voices contributor Keith Lehmann, a technology consultant and former managing director of the Connected Car Council, breaks down the impact of AV and ride-sharing for car ownership...

The big picture: An initial drop in private car purchases will likely be limited to dense cities, and it's still too early to know how it will be affected by other trends in urban planning and public transit.

Background: Vehicles lose 40% of their value to depreciation in year one, and car owners deal with ongoing associated costs — insurance, maintenance, financing, fuel, parking and more.

  • Even so, the people who do the most driving, chiefly in suburban and rural areas, are the least likely to give up their cars, given current trends and more individualized commuting routes.
  • Ride-sharing accounts for only 1% of vehicle miles traveled and has so far had a negligible impact on car ownership.

What's happening: The impact of AVs on private car ownership remains speculative, but their arrival could coincide with increased traffic congestion, decreased vehicle production and improvements in mass transit.

  • It's expected that high initial prices for AVs will make private ownership economically unsustainable, but that ride-sharing services and their ability to monetize their fleets could offset the upfront costs of large-scale deployment.
  • Both Uber and Lyft are also expanding their offerings to include bicycles and scooters. The popularity of bicycles has peaked during past surges in gas prices, suggesting a relationship between car costs and bike use.
  • Some automakers, including Ford, are reducing the number of models they manufacture, and Ford, GM, and Tesla have all taken recent steps to cull their workforces — possibly signaling they see a future with fewer car purchases according to Wards Auto.

What to watch: Public investments in mass transit may be a more reliable indicator than tech investments of whether people will move away from reliance on private vehicles.

  • Increased and improved mobility options won big in last week's midterm elections, when roughly 83% of public transportation ballot initiatives passed, in places from Maine to Florida.

Go deeper: Read the full post.

4. 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 to joann.muller@axios.com.

EV offensive: Volkswagen sets $50 billion target for spending on EV and AV over the coming 5 years. (Edward Taylor — Reuters)

  • Quick take, from Reilly Brennan, founding general partner of Trucks Venture Capital:
"Finally it appears we have an entity that's intellectually honest about the capital allocation required for these tasks."

Doubt: Apple co-founder doesn't believe in self-driving cars (Tyler Clifford — CNBC)

  • Why it matters: Steve Wozniak used to be a believer, but apparently his disappointing experience with Tesla's AutoPilot soured him on the potential for driverless cars.
  • My thought bubble: Instead of using Tesla as the bar, he might want to check out progress at Waymo or GM. Or even at his former company, Apple, which has been rehiring AV engineers lately.

Found: An autonomous rover spots an Argentinian submarine missing for a year (Dave Gershgorn — Quartz)

  • My thought bubble: I wonder if an underwater robot can find my sunglasses that fell overboard last summer.
5. 1 parking thing: robotic space savers
Screenshot from "MyPark in 30 seconds" demo video on YouTube.

Here's a potential time-saver this Black Friday: a smartphone app that will deploy a robot to hold a parking space for you at the mall.

Details: MyPark is a Miami-based parking reservation service that, per Forbes, deploys mobile space blockers to hold spots for customers willing to pay a few dollars for the convenience.

  • The app triggers a pop-up rectangle that blocks other cars from parking in the reserved space.
  • When the rightful renter arrives, it folds down flat so they can drive over it.
  • The startup has deals with some of the country's biggest mall operators, including Simon Property Group and the Mall of America.
  • Its spaces are close to the mall entrances, making the reservation service doubly attractive.

My thought bubble: Parking is a pain, especially during the holidays. If you can use technology to save your spot, you don't have to push your kid out of the car to stand there while you circle back around. And you don't have to tip the valet, either.

Joann Muller