Jun 26, 2019

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.

  • Today's Smart Brevity count: 1,189 words, <5-minute read.
  • Expert Voices contributor Chris Gill writes about why speed is so important to AV simulation.
1 big thing: Truck driving could become a desk job

Illustration: Sarah Grillo/Axios

Last week, for likely the first time, a heavy-duty commercial truck drove for 9.4 miles on the Florida Turnpike with no one inside. The "driver" was 140 miles away, operating the rig remotely.

The big picture: Automated freight delivery is expected to begin long before self-driving cars are here, and at least a half dozen truck companies are working on the technology, with tests in various stages of development.

  • Starsky Robotics' Florida demonstration was believed to be the first unmanned, high-speed test of a heavy-duty commercial truck on a public highway.

Why it matters: The U.S. is experiencing a severe shortage of truck drivers — as many as 175,000 by 2026, according to the American Trucking Associations. Companies like Starsky Robotics hope they can address the shortage by making the jobs less taxing.

"The problem is there aren't enough people willing to spend a month at a time in a truck."
— Stefan Seltz-Axmacher, Starsky Robotics co-founder

Instead of aiming for an AV moonshot — an autonomous truck that makes all the driving decisions without any human intervention — Starsky says it's taking a more practical approach.

  • Its system combines highway automation with teleoperation, allowing remote drivers to navigate trucks between distribution centers and the highway.

Details: With no one inside, the Starsky truck navigated a rest area near Orlando, merged onto the highway from the left, kept a speed of 55 mph, changed lanes, and exited the highway on the right through a toll booth.

  • The remote driver — sitting behind 3 computer screens in an office 2 hours away in Jacksonville — used a steering wheel, buttons and foot pedals to maneuver on and off the highway.
  • After he set the speed to 55 mph, the automation took over, with the driver intervening only to order the lane change.
  • In all, the human driver operated the truck for just 0.2 miles, or 2% of its journey, says Seltz-Axmacher. "It got pretty boring," he says.

Yes, but: Teleoperation relies on ordinary cellular networks that occasionally lead to communication glitches that could potentially delay remote decision-making.

The bottom line: Automated trucking is getting closer, but the instincts and knowledge of human drivers are still needed, even if the humans themselves aren't in the vehicle.

Read the full post.

2. Apple acquires Drive.ai's assets and talent

Photo: Drive.ai

Apple has purchased Drive.ai, an autonomous driving startup once valued at $200 million, and has hired dozens of Drive.ai engineers, Apple confirmed to Axios' Kaveh Waddell and Ina Fried.

The big picture: The deal and hires confirm that Apple hasn't given up its autonomous driving project, though it has changed direction and leadership a couple of times.

What we're hearing: Drive.ai talked with multiple potential acquirers, but in the end Apple won out.

  • Drive.ai ceased operations within the last 2 weeks, and the San Francisco Chronicle reported the company is laying off 90 workers in California.
  • Apple's hires are mostly in engineering and product design, per a source.
  • Apple also purchased Drive.ai's autonomous cars and other assets.

The purchase price was not disclosed, but is likely less than the $77 million Drive.ai raised in venture capital, Axios' Dan Primack reported recently.

The backdrop: Drive.ai's highlighter-orange vans ferried workers around a business park in Frisco, Texas, and shuttled fans in nearby Arlington to Cowboys games.

Separately: Troubled AV maker Faraday Future has reportedly fired several dozen of the employees who have been on unpaid leave, according to The Verge.

Also: Uber is buying Mighty.ai to boost its self-driving car efforts.

3. Why speed matters in AV simulations

Illustration: Rebecca Zisser/Axios

Simulation is a crucial tool in the development of self-driving cars, but it has constraints, especially regarding the speed at which it takes in data and makes decisions, Chris Gill writes for Axios Expert Voices.

Why it matters: In the absence of significant amounts of data from autonomous vehicle road testing, companies like Waymo, Uber and Tesla are investing in self-driving simulations, with billions of miles in virtual simulation tested to date.

  • Speed is critical because if sensing does not keep up with the rate at which events happen, the vehicle might not be able to respond effectively and adapt its driving pattern.
  • If a simulation does not process an event quickly enough, it could lose track of how the consequences of key events affect future behavior, leading to a loss of fidelity in the testing.

Between the lines: In a research setting, advanced cameras in vehicles with automated features typically capture images at a speed of up to 300 frames-per-second, and this speed informs simulations.

  • But cameras in anticipated road models are expected to be slower, which could create a lag time that limits a vehicle's capabilities to identify and respond to an event in near real-time.

The bottom line: Data from varying sources, simulations included, will be crucial to getting AVs safely on the roadways — but simulation technology can improve when it comes to speed, and those improvements will yield better data and stronger AV training.

Read the full story.

Chris Gill is a professor at Washington University's McKelvey School of Engineering. 

4. Driving the conversation

Scoop: Waze and SpotHero tackle parking-related street congestion (Kia Kokalitcheva—Axios)

  • Details: SpotHero is integrating Waze, a navigation app owned by Google, into its app to help customers navigate to a pre-booked parking spot near their destination, Kia reports.
  • Why it matters: Americans spend 17 hours per year on average searching for parking, costing them $345 per driver in wasted time, fuel and emissions, according to INRIX.

Research boost: Argo AI is investing $15 million into a self-driving car research center at CMU (Kirsten Korosec—Techcrunch)

  • The big picture: the investment in CMU, following Argo's release of HD maps and other data to academic researchers, is the latest effort by the Ford-backed company to accelerate the development of self-driving cars.

Driving machines: BMW Unveils Concept Vehicles You Drive, Or Don't (Ed Garsten—Forbes)

  • What to watch: The German automaker seems to be hedging its bets, trying to preserve its reputation for sporty cars that are fun to drive, while also transforming mobility for the future with cars that drive themselves.
5. Crossing the EV profit desert

Photo: Michael Kappeler/picture alliance via Getty Images

Auto industry profits are going to be squeezed for the foreseeable future as companies spend heavily on electric and autonomous vehicles amid stagnating global sales, according to new research from Alix Partners.

Why it matters: The high cost of battery electric vehicles (BEVs) means there is a near-term "profit desert" the industry must cross until consumer demand grows, Alix forecasts.

  • In the meantime, paradoxically, the move toward cleaner cars means automakers will have to keep pushing hard to sell as many high-margin, gas-powered SUVs and pickups as they can.

Details: The global auto industry plans to invest $225 billion on electric vehicles between now and 2023, according to new research previewed Tuesday at a meeting of the Automotive Press Association.

  • The variable cost of an electric powertrain is about $16,000, Alix says, about 2.5 times more than the $6,500 cost of a gasoline powertrain.
  • Batteries are the biggest factor, at about $10,000 per vehicle. But costs per kilowatt hour are coming down, and are expected to fall below $100 by 2023, from 2018's $176.
  • Although consumer interest in EVs is beginning to increase, sales per EV model will be just 14,000 units by 2022, compared to 93,000 average sales per model for gasoline cars, making it hard to amortize those investments.

What to watch: Alix says public charging infrastructure is not keeping pace with the growth in electric vehicles in the U.S., and forecasts a glut of battery production capacity in China will grow over the next decade.

Joann Muller