Chinese Robotaxi Firm AutoX Starts Operation In Shenzen With No Human Driver

Most self-driving car companies move to test vehicles put putting them into taxi service, to learn what it’s like to carry passengers, but they always start by having a human safety driver on board who, ideally, never touches the wheel, though at first they need to intervene from time to time.

As progress is made, an important milestone is to have the confidence (and legal permission) to operate for actual passengers without the safety driver. On Dec 2, AutoX, a Chinese robocar developer, announced that their vehicles are now providing taxi rides in Shenzen to a group of staff and VIP guests.

Google DriveVideo – AutoX Puts Fully Driverless RoboTaxis on the Roads in China.mp4

The start of such operations presumes that the existing tests have shown an excellent safety record, good enough that the company is willing to take the necessary risk and liability of no-driver operation. Previously, Waymo was the only pioneer to do this, having begun in 2018. Since then it has expanded operations and temporarily is making all rides passenger-only due to Covid-19. AutoX is deploying its entire fleet of 25 cars passenger-only, and did not disclose what set of roads the vehicles operate on. (Waymo operates in the fairly easy region around Chandler, Arizona.)

A handful of companies (including AutoX) have permits to do testing with no safety driver in California, but nobody has yet to actually begin operations. Cruise has said it will do so soon. Yandex

YNDX
operates their test taxi fleet in Russia with the safety driver in the passenger seat — which means they can still access an emergency brake switch and even grab the wheel — as an intermediate step to full uncrewed operation.

Recently, Waymo published a safety report to show just how good their safety record had been and why it justified, even 2 years ago, the removal of safety drivers on some rides. AutoX has yet to publish such a report but suggests one is coming soon.

While AutoX indicated the vehicles do not have remote monitors or operators, the vehicles will come to a stop if they come to a confusing situation, and a remote operator can then give high level guidance on how to proceed. This is similar to how many other teams do remote operation, including Waymo, though Waymo has more constant monitoring at present and their operators can act before the vehicle stops, though they don’t do “remote control” style steering. A few companies have made efforts at complete remote control, including AutoX’s Chinese rival Baidu

BIDU
. Nuro, which makes delivery robots which have no room for a driver, has always had to operate them that way.

The AutoX vans have 4 LIDARS and imaging radar. They are the 5th iteration of AutoX’s design. AutoX also has 75 other robotaxis in Shangai, where service is available to members of the public who can hail them via the AliBaba AutoNavi app.

Who’s next?

Most teams today test with two safety drivers. On the way to zero, some may pass through having only one, though the famous fatality when Uber

UBER
did that has left most teams with two. California has allowed several companies to get permits to test with no safety driver, but none have taken that step.

California permits belong to Cruise, Nuro, Waymo, Zoox and AutoX. Waymo has just greatly expanded its San Francisco facilities. Zoox has teased a big PR announcement coming up on Dec 14. AutoX has not indicated they will use their permit in California soon.

Of course, Tesla

TSLA
continues to test their system with untrained customers, allowing them to rack up far more miles than any other team, though with less data from these miles.

California’s public utilities commission also recently declared it will allow robotaxi companies to charge money for rides — but only after a long and probably arduous regulatory process. It’s not clear why they want to be so slow about allowing charging. All the companies doing ride services (with safety drivers or eventually not) don’t want to charge to make money — not yet — but they do want to experiment with business models, and see how customers react to different pricing approaches. The CPUC is protecting nobody by slowing this down. It’s not clear why they should regulate prices at all. Yes, when it happens it’s going to outcompete human taxi and Uber drivers, but it will also outdo them in safety, and it’s tilting at windmills to stop that process.