China’s top scholars are calling for more policies which encourage data sharing and product standards for autonomous vehicles, and advocating for higher levels of autonomy for testing the technology.
“Large-scale production should only be for vehicles meeting the Level 4 requirements, while Level 3, which involves transferring control from car to human cars, should only be applied to research,” (our translation) Li Deyi, a Chinese Academy of Engineering (CAE) fellow, said Friday at this year’s World Artificial Intelligence Conference (WAIC) in Shanghai.
Level 4 (L4) autonomy refers to a fully autonomous system which can handle emergency situations. L3 still requires that a driver intervene in emergency cases, according to definitions set by the Society of Automotive Engineers (SAE).
Li’s comment echoes a long-held debate in the industry over whether such handovers are safe for owners. A number of tech giants and automakers argue that a machine should assume full responsibility, including Alphabet’s Waymo, Volvo, and trucking unicorn Tusimple. Others favor a more realistic technology approach for semi-autonomous cars. Chinese automakers GAC Group, Changan, and XPeng Motors plan to produce L3 automated vehicles by next year.
“In China, the public cares more about safety, and so the current problem for cars testing on the road is, what are the safety requirements that should be met?” Li asked. He proposed that the government release safety standards—such as the allowable scope of failure rate and specific autonomy levels for cars permitted to conduct trial runs on public highways—as early as possible to accelerate commercial development for the industry.
Legislation for data management is another pressing need in China’s self-driving industry, experts at the conference said. China needs to formulate a set of unified rules for data processing, transmitting, and sharing, none of which exists under current national cybersecurity laws, said Wang Yao, director of technology at the China Association of Automobile Manufacturers (CAAM).
Data is considered immensely valuable for developing autonomous vehicles and has become one of the key issues between automakers and tech companies as both sides fight for control. Alibaba, an exclusive partner to SAIC for vehicle operating systems, is barred from accessing most of the state-backed auto giant’s driving data, according to Caixin.
The lack of collaboration points to insecurity, because “automakers are under great pressure as internet giants penetrate the industry,” Wang explained. He added that the Chinese government has started refining the country’s cybersecurity law to build explicit rules for auto-related data, such as car location data, surrounding data, and engine state information to encourage industrial collaboration.
“We hope Chinese automakers will form alliances first to build data-sharing platforms,” Wang said.