Waymo has been offering its Robotaxi service in an area of 260 square kilometers in the Phoenix metropolitan area since 2018. Since 2020, anyone with a smartphone can order a robot taxi at any time. Neither is there any need for a declaration of confidentiality, as is the case with Waymo One’s early rider program.
In the course of the corona pandemic – as with the competition – all test drives on public roads were initially discontinued. But Waymo has been back at the start since September 2020, albeit with a different strategy: The safety driver has disappeared! This was still on board as standard in order to meet the subjective safety needs of the passengers (see The “Chauffeur” project is starting).
The pandemic is accelerating development
In the course of the corona pandemic, this has apparently changed. Conversely, the fact that nobody is on board is a convincing argument for such a service from the perspective of infection protection. The legally prescribed possibility of intervention in the case of disengagement is satisfied by remote monitoring.
Autonomous driving, the effects on the transport of people with robotic taxis, people movers, autonomous buses and the necessary techniques and regulations deal with us in a ten-part series, the individual articles of which will be published on weekdays until May 14th.
- Part 4: USA – Eldorado for test operations with robotic taxis
Test operations are running in some states, and California has even regulated commercial operations. The industry is hoping for a breakthrough from Joe Biden.
- Part 5: Robo-Taxis – Who Will Win the Race in the USA?
Two dozen companies are testing autonomous fleets, and the race will likely be decided in California. We give an overview of the applicant field.
- Part 6: Like California, only bigger: Robotaxis in China
The potentially largest market for autonomous fleets is impressive in terms of technology, test operations and capital resources of the companies involved.
- Part 7: China: Autonomous driving through a planned economy
Startups that are keen to experiment, state infrastructure funding and a high level of acceptance among the population make autonomous fleets a recipe for success in China.
- Part 8: Automatic Shuttles – The European Way?
There are no robotaxi test runs in Europe, but more and more autonomous shuttles that, as part of public transport, undertake cautious driving tests …
- Part 9: Green wave for robotic taxis in Germany too?[br]These days the Bundestag is debating a legislative proposal from the Ministry of Transport that could pave the way for robotic taxis in our country too.
- Part 10: Driverless transport as part of the traffic transition
Robo-taxis, people movers and autonomous shuttles – can autonomous fleets make a contribution to the traffic transition?
As soon as the vehicle is unable to cope with a situation, a Robotaxi pilot who sits in a remote control center takes over. The traveling one Security personnel can thus be savedwhich eliminates a significant cost factor of the business model.
Waymo: Safety first, vision zero
According to the company’s own statements, safety is a top priority at Waymo, in September 2020 the “Waymo Safety Report” presented to the public. Between 2019 and the end of September 2020 there were 18 minor accidents, 16 of which were rear-end collisions with other vehicles. Only once did a Waymo vehicle run up when another car abruptly changed lanes and braked sharply. In three cases, the car was bumped into the stand by cyclists or pedestrians, and in two of these cases the vehicle was rammed by cars moving backwards. Once there was a head-on collision when the opposing vehicle was on the wrong lane.
There were no serious injuries. “Almost all events … related to one or more violations of traffic rules or deviating driving behavior by the other road user”, it says in the report.
Waymos vehicles travel at a maximum of 45 miles per hour (approx. 72 km / h), but do not drive on the motorway, although both the vehicles and the operating license would allow it. The vehicles also drive in the rain and at night. The redundant sensors, especially radar, also work well in rain, snow and fog, the technology is far less sensitive to environmental influences than human drivers are. Nevertheless, raindrops or snowflakes can impair the sensor function. There have been reports of Waymo cars stopping briefly and waiting in heavy rain.
Waymo tests vehicles at 25 locations in more than ten US states. Over 20 million miles on public roads and more than 15 billion in the simulator have been covered. Up until the beginning of 2020, between 1,000 and 2,000 Waymo One journeys were made per week, 5% to 10% of which were completely driverless (Waymo One, Waymo Safety Report, September 2020). For Phoenix, Arizona, the primary test location for Waymo, the report shows 6.1 million miles driven. Without a safety driver on board, a total of 65,000 miles had been driven in Phoenix by September 2019.
In view of the many miles driven, the accident rate is very low. In a direct comparison, the susceptibility of human drivers to errors is noticeable. Apparently, non-autonomous accidents are taken for granted. Every 280,000 kilometers on average someone causes a car accidentreported to auto insurers (2019). None of the accidents that Waymo lists in its report fall into this category, as they were all due to third-party negligence ().
Acceptance: Zero Tolerance
Waymo has come to the conclusion that when the technology is introduced, a very high level of sensitivity is to be expected with regard to the susceptibility to errors and that acceptance will only be given if the number of accidents remains minimal. Observers see a change in strategy here. Simply being safer or significantly safer than human drivers obviously not enough.
This view has been supported by reports of a few fatal accidents involving Tesla and Uber. These received worldwide attention and led to waves of criticism against the respective companies. In the case of Uber, one of the consequences was the temporary cessation of operations with automated driving functions.
The first head of the Google Car Project, Chris Urmson, emphasized the crucial difference between “continuously improved driver assistance systems” and “completely driverless systems” at a legendary TED talk back in 2015. He drew attention to the problem of unfocused drivers who overestimated the capabilities of the assistance systems. From early experiences with unfocused test drivers on the autobahn, the company drew the conclusion that it would no longer pursue this development path.