Weekly Brief: Games Accident Shows Even Giants Stumble in an AV World

A Toyota self-driving shuttle operated by a human driver struck a visually impaired athlete at the Tokyo Paralympic Games last week.

The accident happened on just the second day of the two-week event, where Toyota was hoping to highlight the safety, efficiency and potential of its autonomous, all electric e-Palette transportation pods. Toyota had spent months specially designing the pods in collaboration with Paralympic athletes. They featured electric ramps for riders in wheelchairs and handrails and seats that accommodated passengers of all heights and physical abilities. The carmaker loaded the pods with autonomous tech that was supposed to be able to handle routes in and around the bustling Athlete Village.

The pod was traveling between 1-3kmh when it struck Japanese Paralympic judo athlete Arimitsu Kitazono, who was crossing a crosswalk. Kitazono was treated in the Athlete Village for head and leg injuries and missed his scheduled event. Toyota suspended its entire fleet the same day, as the company and the police opened investigations. “It shows that autonomous vehicles are not yet realistic for normal roads,” said Toyota CEO Akio Toyoda.

To be fair, an accident involving an autonomous vehicle operated by a human driver is different than an accident involving an autonomous vehicle operating in autonomous mode. When a safety operator switches an autonomous vehicle into manual mode, the autonomous tech defers to the human’s judgement, which in this case reasoned that the athlete in the crosswalk would yield. Would the autonomous tech have behaved differently on its own?

It could be telling that the safety operator was driving in manual mode, given possible concerns over the tech’s inability to handle the busy environment but that’s conjecture. We’ll have to wait for more details before we can draw any specific conclusions about who or what was at fault here. Either way, it’s yet another setback for a self-driving car industry that’s starting to feel like a walking cautionary tale about human hubris in the machine age.

Robo-taxi services that were supposed to be up, running and commercialized years ago now, or so according to overzealous CEOs talking to overzealous reporters (myself included), are still inching along in small-scale pilots, years away from commercial deployment let alone profitability, all the while sucking down hundreds of millions of dollars in venture capital.

Consider Toyota. In 2018 it invested $2.8Bn to create a new company called the Toyota Research Institute-Advanced Development (or TRI-AD), with the stated goal of deploying autonomous vehicles by 2020. That deadline sailed by. In April 2021 Toyota invested an additional $550M to acquire Lyft’s self-driving outfit, Level 5, after Lyft had grown disillusioned with its own dream of deploying robo-taxis across the US. All of that investment has led to this, an e-Palette shuttle that still needs to be manually operated and gets suspended on its second day in operation.

No wonder the self-driving car industry is contracting so rapidly. Fewer and fewer players want to take the risk, or can afford to, and only the largest, most well-funded outfits can sustain themselves. All of which is music to the ears of Waymo, which has arguably the best, most rigorously tested technology in the world, one of the only commercial deployments and with Alphabet’s deep pockets to keep it funded.

Last week it hit an important milestone in San Francisco, where local residents can now sign up to ride in its self-driving vehicles for free in exchange for their feedback. The Waymo One Trusted Tester Program will feature all-electric Jaguar I-Paces loaded with Waymo LiDAR, cameras and radar. The vehicles will have safety operators on board but will operate without their influence, except in emergencies. The trial is open on a first-come, first-serve basis through an app to any San Francisco resident who is vaccinated and Covid free.

The operator did not specify how long the trial will last or how soon it could transition to a fully commercialized deployment but with Phoenix as a guide, don’t expect anything soon. From the company’s perspective, why rush? It’s playing the long game and appears to be winning.

One comment

  1. Avatar Timo 30th August 2021 @ 5:35 pm

    It seems to me that the human in control of the vehicle is what caused the vehicle to hit the person. If the vehicle is being controlled manually, it has nothing to do with autonomy. If no collisions occurred with these vehicles in autonomous modes, and a collision occurred with a vehicle being controlled manually, does is not make the case that the autonomous vehicle is safer?

Leave a comment

Your email address will not be published. Required fields are marked *