News from the Real World of Self-Driving Taxis: Not YetWayMo includes a human in all their “robotaxis,” just in case, because the vehicles (at last report) were still confounded by common conditions
The typical “snippet sentence” news we get about AI says a little and then lets the imagination zoom. How about this item from Forbes: “Google and Tesla Say Self-Driving Tech Will Accelerate Exponentially Within 24 Months,” featuring both Elon Musk of Tesla and Jack Krafcik, CEO of Google-owned WayMo.
You must look below two photos, lots of hype, and three links leading to other stories of interest before you encounter the industry skepticism due to technical challenges. That same day (April 24, 2019), we were told that Tesla missed big on its first quarter, due to falling demand for its electric cars. Many sources, including Jonathan Bartlett here at Mind Matters News, suspect that the froth also hides technical gaps that are way bigger than advertised.
Here are some gaps to think about. The National Highway Traffic Safety Administration identifies six levels of driving automation. Level 0 is no automation at all; Level 5 is no human at all. While the levels between 0 and 5 shift the balance from human to automation, the shifts are not smooth and gradual. There are significant gaps between some levels.
For example, consider Level 4: “The vehicle is capable of performing all driving functions under certain conditions. The driver may have the option to control the vehicle.” Judging from all the hype, you would expect “capable of performing all driving functions” to be nearly here. But it comes with the squishy qualification of “under certain conditions.”
A self-driving vehicle could satisfy Level 4 if the roads were equipped with in-ground sensors and all conflicting traffic were removed. Car and Driver Magazine explains, “But in a privately-owned Level 4 car, the driver might manage all driving duties on surface streets and then become a passenger as the car enters a highway.”
While Musk’s announcement appears to be yet more Tesla grandstanding (there’s a long history), WayMo has provided details we can evaluate. The firm intends to lease a factory in Detroit where they will retrofit Jaguars and Chrysler minivans with their technology. They plan to deploy those vehicles for a “robotaxi” service in southeast Michigan; much like that which they’ve deployed in Arizona.
These details both reveal the challenge and temper the hope of “exponential” change. Arizona, and especially the Phoenix area, is well-suited for autonomous vehicles: It gets lots of sun and little rain—so the sensors get a clearer picture. It is flat with relatively new streets. Thus it is much easier to navigate than, say, Boston or Manhattan. And, even with those caveats, WayMo includes a human in all their “robotaxis,” just in case, because the vehicles (at last report) were still confounded by common conditions:
Inclement weather, and rain specifically, poses another problem. Metz said he hailed a Waymo vehicle during a recent rainstorm, and when he got in, discovered that the company’s human monitor was driving the car manually. But Metz said he doesn’t fault Waymo for being overly cautious, nor does he sympathize with Phoenix residents who have complained about getting stuck behind a slow-moving self-driving car. “If you’re tailgating and doing ten over [the speed limit], then yeah, you’re going to be angry,” he said.Andrew J. Hawkins, “We spoke to a Waymo One customer about how robot taxis get confused by rainstorms” at The Verge
Southeast Michigan is not sunny, not dry, and only kinda flat.
WayMo’s experience in Phoenix “reveals a brutal truth”:
So no, this is not the anyone-can-ride, let-the-robot-drive experience Waymo and its competitors have been promising for years. Building a reliably safe system has proven far harder than just about everyone anticipated and its cars aren’t ready to drive without human oversight. But Waymo promised to launch a commercial service sometime in 2018, it didn’t want to miss its deadline and risk its reputation as the leader of the industry it essentially created, and not even the might of Waymo parent company Alphabet can delay the end of the calendar year.
So Waymo is pushing out a software update, tweaking its branding, and calling it a launch…
“We’re definitely under no illusions about the path ahead of us,” says [WayMo Product Chief Dan] Chu, “I think anyone who says otherwise doesn’t understand the challenges as deeply as we do.”Alex Davies, “Waymo’s so-called robo-taxi launch reveals a brutal truth ” at Wired
The hype surrounding autonomous vehicles from nearly all of the AI media arises from overconfidence in technology and underappreciation of humans. For example, NHSTA states that 94% of all traffic accidents involve human error. But then, humans drive 100% of the cars and roads are not known to frequently leap up and cause accidents. So, what’s left? Human error.
It’s wildly presumptuous to state that autonomous vehicles will fare better when even simple, everyday situations confuse them. The best experts believe that removing the driver is, at best, decades away. And, we are told, Elon Musk and his ilk are seriously annoying them.
Hype serves no one other than the early investors hoping to get their cash back. Calm evaluations—and an appreciation for the amazing beings that humans are—would serve all of us much better than overpromised claims that are doomed to under deliver.
Also by Brendan Dixon: The superintelligent AI myth
See also: Are Tesla’s robot taxis a phantom fleet?
Can we program morality into a self-driving car
Note: The image above is Phoenix, Arizona, as seen from South Mountain, courtesy DGustafson (publish domain)