Self-Driving Truck Engineer Makes Revealing Admission as Company Ramps up Testing
Mountain View, CA – Another self-driving technology executive is voicing just how difficult it is to achieve the goal of putting unmanned trucks on American roadways in a commercial application.
In a press briefing on Tuesday, Boris Sofman, Waymo’s lead truck engineer, admitted achieving a driverless truck in a real-world environment has proven to be much more difficult than many originally thought.
“The difficulty of actually getting to a point where you can move from a very compelling and reliable demo with a safety driver to actually having the confidence to pull a safety driver out of the seat and take real goods on a public road in fully autonomous mode was way bigger than I think anybody expected,” Sofman said.
His admission comes as Waymo, the self-driving technology company that’s part of Alphabet (Google’s parent company), is ramping up testing in the American Southwest.
Much like its competitor TuSimple, Waymo is focused on running tests primarily along Interstates 10, 20, and 45 and through metropolitan areas like El Paso, Dallas, and Houston.
Earlier this year, Waymo announced it was mapping routes between Phoenix, El Paso, Dallas, and Houston and ramped up testing in California on freeways in Mountain View.
It also has been in a partnership with the United Parcel Service (UPS) since February of this year.
Waymo’s autonomous Chrysler Pacifica minivans are shuttling packages from UPS Stores in the Metro Phoenix area to the UPS Tempe, AZ hub, as part of a pilot program.
The vehicle operates autonomously with a Waymo-trained driver on board to monitor operations.
Still, Sofman’s revealing comments echo similar sentiments made back in March of this year by former CEO of the now-defunct Starsky Robotics, Stefan Seltz-Axmacher.
Seltz-Axmacher made global headlines when he penned a blog post to announce Starsky’s demise.
In it, Seltz-Axmacher said he doesn’t believe any existing autonomous truck startups will ship vehicles with artificial intelligence (AI) capable of performing to an “human equivalence” standard within the next ten years.
“The biggest limiter of autonomous deployments isn’t sales, it’s safety,” he declared. “By definition building safety is building the unexceptional; you’re specifically trying to make a system which works without exception.”
To be clear, it’s not that self-driving technology can’t operate big rigs safely in certain “limited-use cases,” because it can, Seltz-Axmacher argued.
The problem is that building a driverless truck with AI that never fails is “really, really hard,” he conceded.
The question many are asking is, how much error and risk are investors, stakeholders, policy-makers, and the public willing to accept in pursuit of realizing the goal of deploying unmanned trucks on U.S. roadways?
Only time will tell.