Standards to make autonomous vehicles safe

Editor's note: China has released the first mandatory national safety standards for Level 3 and Level 4 autonomous vehicles. China Central Television spoke to Xiong Lu, dean of the College of Automotive and Energy Engineering at Tongji University, to explain the standards. Below are excerpts of the interview. The views don't necessarily represent those of China Daily.

The new standards establish a unified and mandatory safety threshold for entry into the market, providing a clear basis for the future development of autonomous driving systems. They specify quantitative standards, parameters and technical requirements for products and establish comprehensive safety requirements for the entire life cycle of a vehicle. They are mandatory standards and all vehicles equipped with autonomous driving functions must comply with them before they can operate on public roads.

At the system level, the new standards set clear quantitative thresholds in two key areas. First, they mandate that the driving capabilities of autonomous driving systems must be comparable to those of a qualified and attentive human driver. Second, the systems must attain an accident rate comparable to conventional vehicles, which is roughly one accident in every 10,000 hours of driving. This is a key technical safety threshold established by the standards.

From a research and development perspective, two requirements are particularly important. For Level 3 autonomous vehicles about to be mass-produced, the vehicles must have a driver takeover system. Additionally, in hazardous situations that can't be managed by the system, the standards require a 10-second window for the driver to take control. These two specific requirements have received considerable attention within the R&D community in the industry.

The standards mandate a comprehensive safety assurance mechanism throughout the entire life cycle of the vehicle, encompassing initial R&D, production, supply chain management, testing, evaluation and data monitoring during operation. This comprehensive approach is important because autonomous driving systems are very complex, and it becomes challenging to determine responsibility in accidents. Only by managing the entire process across the enterprise can responsibilities be clearly determined.

For many companies, establishing a comprehensive life-cycle safety assurance mechanism will be challenging. The requirements cover the entire R&D process, necessitating complete procedural and management systems.

Manufacturers that rely on third-party suppliers need to collaborate closely with those entities to establish the necessary life-cycle management procedures and standards across the entire system. Coordination with component suppliers and regulatory authorities may also be necessary, making the implementation of these requirements a demanding task.

The standards will evolve with additional technical criteria. Further refinements will address functional safety, intended functionality safety and cybersecurity. There will also be corresponding standards for perception systems, vehicle connectivity systems and other areas. Additional standards for the testing and evaluation of autonomous driving systems will also be established. Also in the pipeline are standards for the overall architecture of autonomous driving systems, as well as for operating systems and chips.