Driverless cars may cut crashes, but computers do fail

PUBLISHED : Saturday, 14 October, 2017, 9:00am
UPDATED : Saturday, 14 October, 2017, 9:44pm

I cannot agree more with Wendell Chan, of Friends of the Earth, that policymakers have to think seriously about the role of self-driving cars in Hong Kong and how the city will adapt to it (“Self-driving cars in Hong Kong may lose their edge with selfish owners”, October 8).

In fact, a recent study found that self-driving cars hold the potential to transform driving by eliminating the majority of traffic deaths, significantly reducing congestion and providing tens of billions of dollars in economic benefits.

If only 10 per cent of cars on the road were self-driving, the number of traffic deaths could be cut by 10,000 a year and these cars would produce nearly $38 billion (HK$296 billion) in economic and other savings, said the study by the Eno Centre for Transportation, a US foundation dedicated to improving transportation.

If 90 per cent of vehicles were self-driving, as many as 21,700 lives could be saved each year, and economic and other benefits could reach a staggering $447 billion, said the study.

Government research in the US has shown that driver error is most likely the main reason behind 90 per cent of all crashes. More than 40 per cent of fatal traffic crashes involve alcohol, distraction, drugs or fatigue. However, with self-driven vehicles such human failings don’t have a potentially deadly impact, suggesting at least a 40 per cent reduction in fatal crashes could be possible, the study said.

Crashes can also be due to speeding, aggressive driving, over-compensation, inexperience, slow reaction times, inattention and various other human driver shortcomings, the report noted, suggesting the computer doing the driving could also cut those.

There are disadvantages, though. In an accident, who would be liable – the person behind the wheel or the maker of the software?

Self-driving cars are just like robots, and there would be times the machine will malfunction due to wear and tear, and overwork. Under such circumstances, would the machine be able to give human drivers warnings that they are not fit to do the job any more?

Eunice Li Dan-yue, Shanghai