網上版請按此
Automated driving era: we may have to wait a bit longer
Today, all the world's major automobile manufacturers, such as BMW, Mercedes-Benz, Volvo, Audi, Honda, and Toyota, and even the digital giants, including Apple, Baidu, Microsoft, Samsung, etc., have indicated their intention to launch autonomous vehicles (AVs).
However, they differ in their definition of "automation". The SAE International Association lists six levels of automated driving according to the level of human participation: the highest three levels of full automation are without human monitoring, such as the level of "high automation" where the driver can sleep while traveling.
We have heard of AVs for a long time, but when can we actually have it? It is said that Tencent will first market a small quantity of AVs by the end of 2018 or early 2019. Several other automobile manufacturers pledge to launch the vehicles (including models without a steering wheel or pedal) before 2021.
In April, Waymo, the Google self-driving spinoff, invited hundreds of residents in Phoenix, Arizona, to try its self-driving vehicles. The test area is equivalent to two San Franciscos, or one and a half times as big as Hong Kong Island.
Over the past eight years, Waymo has accumulated more than 3 million kilometers of actual driving record on public roads. However, according to Car Advice, an online platform for editorial reviews of new cars, the data as of the end of 2016 showed that on average Waymo cars need human intervention every 8,000 kilometers of driving in order to prevent an incident.
The figure is higher than the average accident statistics in the United States. According to the US Department of Transportation, traffic accidents killed 35,000 people and injured more than 2 million in 2015. The figures seem astonishing, but do not forget the average annual mileage of Americans is 300 billion, which means that the accident rate is actually 1.12 deaths and 76 injuries per 100 million miles. By comparison, Waymo's safety record is not promising.
Global consulting firm McKinsey interviewed over 30 European, American and Asian experts to come up with a more realistic timetable. And based on the survey, it concluded that from now until the mid-2020s, AVs will not be available to consumers.
Priority will be given to industrial fleets, such as those for mining and agriculture, which drive within a relatively well-defined environment; then to trucks on highways, and then to the commercial fleets that deliver mail.
It is estimated that by mid to late 2020s, the AVs will be introduced to the consumer market. It takes several years more for the AVs to become popular.
This estimation is similar to the customer survey conducted this year by Gartner, another consulting firm which interviewed 1,500 Americans and Germans.
More than half of the respondents expressed concern about the safety of AVs. With limited demand at the early phase, the price of each AV can be as high as US$100,000 (HK$780,000) initially. Also, the road authorities would take years to develop relevant regulations.
These factors make us believe that automated driving could only be used by the public by 2025 or after with various promotions offered by the automobile manufacturers.
Although it takes more than a decade or more for this mode of driving to go mainstream, we could imagine that the AVs will have a significant impact on our everyday life. One of which may be more time for leisure and entertainment.
McKinsey anticipated that the time saved in commuting with AVs would be as much as one billion hours per day across the world: that's enough to build two Great Pyramids of Giza! And the value of such extra time is enormous. It is estimated that every extra one minute of leisure per person a day could bring about over US$6 billion in revenue for the world’s digital media.
In addition, air quality would be improved as the AVs will be automatically optimized, carbon dioxide emissions are expected to be reduced by 60 percent.
The most important benefit is that accident will be substantially reduced to 10 percent of the current rate. In the US, a fatal traffic accident had caused one dead, eight hospitalized, and 100 treated and released from emergency rooms. In 2012, traffic accidents cost the country US$212 billion. With the decrease in the number of accidents, the US can save about US$190 billion a year.
Google previously applied for a patent to an automatic adjustment of the vehicle's impact characteristic – in a crash, if a hard object like a car is detected at the instance of impact, the vehicle's body will become stiff; if it is a human, the vehicle's body will loosen up to ease the impact of the crash.
Later, we may sigh with relief for having less misery from traffic accidents.
Dr. Winnie Tang
Honorary Professor, Department of Computer Science, The University of Hong Kong