At the final day of the Automated Vehicle Symposium 2018 in San Francisco, we moved well past the tech expo phase of the conference and hard into policy. Aside from a panel on venture capital in the automotive technology sector, nearly every speaker represented a government and was discussing either regulatory frameworks or strategic roadmaps for AV testing and implementation.
NHTSA Deputy Administrator Heidi King, who is serving as acting administrator while awaiting Senate confirmation of her nomination to the post, led off the morning with a keynote effectively reeling off the purported benefits of the technology. I don’t think I’ve ever heard anyone jam the word “safety” into a 15-minute speech as many times as she did there, signaling that for all the other value propositions, safety remains far and away the top priority. She also stressed the voluntary nature of AV safety self-assessments, praising Waymo and GM Cruise (though not by name) for having filed theirs, and other AV developers for signaling to NHTSA that they would be submitting their own.
Reading between the lines a bit, it sounded like NHTSA is ready to pull the plug on its DSRC mandate for vehicle-to-vehicle communications, even if it remains committed to using the 5.9 GHz band of spectrum for automotive safety. She suggested the agency would not be picking between winners and losers in terms of telecommunications standards, leaving the door open for utilizing both DSRC and the telco-preferred alternative of cellular vehicle-to-everything communications (C-V2X).
A panel of venture capital executives discussed growing investor interest in automotive technology, which several years ago was on nobody’s radar. One panelist noted the larger auto firms have significant resources, but lack the technology or internal levers to push it forward. Another panelist, Jim DiSanto of Motus Ventures, argued software is hitting a brick wall in terms of development, and suggested the tech industry would be shifting investment towards hardware for the first time in 2-3 decades. He cited the breakdown of Moore’s Law, which has tremendous implications for automated vehicles, given their onboard computing requirements.
The conference then closed with a series of presentations from various governments on the progress in their AV programs, starting with Japan, then moving onto the European Union, the United Kingdom, the United Arab Emirates, and finally the United States.
Some final thoughts on the conference as a whole:
- The international diversity of the conference was impressive, with half a dozen countries sending at least 20 representatives. Overall, attendance was up ~14% from last year’s record, and has tripled since 2014. However, it will be interesting to see the impact of moving the symposium to Orlando for next year’s edition.
- Not terribly surprising, but we saw very little from Waymo or GM Cruise. Waymo had a booth in the expo, and CEO John Krafcik reportedly made a brief appearance, but didn’t speak at any session. Uber was absent entirely.
- The best presentation came from Lyft’s Nadeem Sheikh, which I covered in my first entry of this series. There is also this article on the vision he presented.
- While we were in San Francisco, my friends Ed Niedermeyer, Kirsten Korosec and Alex Roy over at the Autonocast recorded an excellent episode with Comma.ai founder George Hotz. I happen to share his view on the pessimistic outlook for “robotaxis” in the short-term, and the challenges Sheikh outlined fit nicely into that thesis.
- I’m also growing more bearish on highway use cases for high levels of automation in the short-term, even though I’m impressed with startups like Peloton, Embark and Starsky Robotics in the AV freight space. This is starting to feel more and more like a gradual and messy transition due to traffic flow mechanics in mixed traffic. Dedicated lanes are probably 10-15 years out still.
Hope to see another great crowd next year, and I’m looking forward to seeing how far we’ve come!