Why self-driving cars won't be on roads anytime soon

The technology carries a hefty price tag in terms of money, safety, jobs and convincing the public

Pictured here is a self-driving truck made by US company TuSimple. (provided by TuSimple)

There’s growing skepticism about the practicability of autonomous driving technology, which is supposed to let automobiles drive themselves without a human behind the wheel. Despite the rosy prospects offered by the IT industry, which has suggested the technology is right around the corner, it appears we have a long way to go before self-driving cars are cruising the streets.

That skepticism was conspicuous in a forum organized on Tuesday by the Korea Insurance Research Institute and the College of Business Administration at Seoul National University (SNU).

All participants in the forum — which tackled the theme of “insurance and changes in the mobility industry” — raised doubts about whether autonomous driving technology is practicable.

“Realistically speaking, I’m not sure when it will be possible to reach Level 4 or Level 5 of autonomous driving,” said Seog S. Hun, a professor at the SNU business school.

“Level 5 of autonomous driving is still a really long way off. Level 4 has to be conceived in connection with building infrastructure on the streets and elsewhere,” agreed Yi Kyong-su, a professor at SNU’s School of Mechanical and Aerospace Engineering.

Yi is one of Korea’s leading researchers on autonomous vehicles.

The Society of Automation Engineers divides autonomous driving technology into six levels altogether. Level 0 refers to manual driving. Levels 1-2 are when the vehicle assists the driver on expressways and in other specific environments. Self-driving — in which the vehicle can assess the traffic situation on its own — begins at Level 3, which refers to conditional autonomy. Levels 4-5 refer to autonomous driving in the fullest sense of the word, where the driver’s involvement isn’t required.

Even US electric vehicle manufacturer Tesla, the company regarded as having the most advanced autonomous driving technology, is only thought to have reached Level 2.5-3.

Hyundai Motor Company hasn’t succeeded at developing Level 3 autonomous driving technology for use on expressways.

After Google first began work on a self-driving car in 2009, various companies, universities, and governments have invested in the area. But 10 years later, we’re still not even halfway there.

There are several reasons why experts don’t see Levels 4-5 of autonomous driving technology on the immediate horizon. First is the immense cost.

“The cost of building the infrastructure for autonomous driving — including smart roads and mobile networks — would be astronomical. That’s why I think the era of completely autonomous vehicles will come very late,” said Kim Il-pyeong, head of the automobile insurance strategy team at Samsung Fire and Marine Insurance.

Implementing autonomous driving technology would require the control of numerous variables on the road. Safe implementation of the technology will require building roads that are customized for self-driving cars. And most of the vehicles on the road will need to be self-driving cars that minimize human involvement.

Just developing the technology is very expensive. British-based business newspaper the Financial Times reported that Waymo, an affiliate of Google in charge of developing autonomous driving technology, recently raised an additional US$2.5 billion in capital.

The company needed another cash infusion after receiving US$3.2 billion in investment early last year.

Another major challenge is negotiating conflict with competing interests.

“If self-driving cars are adopted, millions of vehicle operators, including taxi drivers, could lose their jobs. As the Tada controversy shows, there are doubts about whether self-driving cars would actually be accepted in Korean society,” said Kim.

Tada, an SME that offered a popular van-hailing service, was forced to shut down under pressure from Korea’s taxi lobby.

“I think that converting a single self-driving car into a shared car would have the effect of replacing 15 ordinary vehicles. It wouldn’t be easy for automakers to convert their business model away from the retail sales market.”

In other words, pushback from workers who depend upon driving for their livelihood and the potential loss of sales for automakers could be stumbling blocks for the widespread adoption of autonomous driving technology.

It would also take time to update legislation and adopt technology.

In 2017, the US government drew up a bill to boldly slash regulations on self-driving cars, and it looked as if the bill was on the verge of passing. But it’s still pending,” said Park Jun-hwan, a legislation researcher with the National Assembly Research Service.

“One reason is that the technology hasn’t achieved the necessary safety and reliability. But more fundamentally, there’s still no solution to the issue of liability for accidents related to data or communications security in self-driving cars.”

“Even if autonomous driving technology is developed in the near future, it will take much longer to build the infrastructure that would make the technology practicable and to convince people — who are the end-users — to accept it,” Jung Sun-young, an economist at the Economic Research Institute of the Bank of Korea, told the Hankyoreh in a telephone interview.

According to the Victoria Transport Policy Institute, in Canada, it took 25 years from the development of the first airbag before they became ubiquitous in vehicles. That suggests just how long it might take before society accepts the novel technology of autonomous driving.

By Park Jong-o, staff reporter

Please direct comments or questions to [english@hani.co.kr]