Intel helps IEEE develop safety decision-making model for autonomous vehicles

Jack Weast, Intel senior principal engineer and vice president of autonomous vehicle standards at Intel subsidiary Mobileye, delivers a keynote speech at the 2019 Mobileye Investor Summit

In order to improve the safety of autonomous vehicles (AVs), the Institute of Electrical and Electronics Engineers (IEEE) has passed a proposal to advance the development of standards for autonomous vehicle decision-making systems. IEEE has also appointed Jack Weast, a senior principal engineer at Intel, as the head of the working group. The working group will be open to the autonomous vehicle industry, and Jack Weast hopes that more industry representatives will join. It is reported that the working group will hold its first round of meetings in the first quarter of 2020.

 “The upcoming IEEE standard will provide a practical tool to clarify the importance of safety for autonomous vehicles.”

——Jack Weast, Intel senior principal engineer

The industry and regulators are working hard to find a unified standard to regulate and ensure the safety of self-driving cars. Although most people support the establishment of an entry threshold for self-driving car driver's licenses, there is still a lack of general consensus in the industry, so multiple standards and regulatory schemes are still in the works.

The constant change and rapid development of technology has brought pressure to the relevant IEEE standard setters, and also prompted them to speed up the formulation of global autonomous vehicle regulations. As we all know, IEEE plays an important role in the field of industry standard setting. In the face of growing calls for regulation, IEEE will promote the formulation of autonomous driving industry standards and is expected to release the first version of the autonomous driving industry standard next year.

“This standardization project will lay an important foundation for the development of open, formal decision-making models for autonomous vehicles,” said Riccardo Mariani, vice president of standards activities at the IEEE Computer Society and vice president of industry safety at NVIDIA . “Redundancy and diversity are critical to developing a scalable safety framework for autonomous driving.”

The new standard "IEEE 2846" will establish a reasonable and standardized mathematical model for autonomous vehicle decision-making. The model embodies the concept of human safe driving into a verifiable model with logically verifiable rules. It is technically neutral and compatible with any set of autonomous driving solutions, and can be adopted by the entire industry. The model can also be customized for different regions. In addition, in order to evaluate whether the model meets the standard, the standard will also include a set of test methods and necessary tools for verifying autonomous vehicles.

The proposal is supported by the IEEE Computer Society and the Automotive Technology Society, both of which are part of the IEEE. Working group chair Jack Weast said the group would welcome participation from "all those interested in developing this important autonomous driving standard."

Intel's Responsibility-Sensitive Safety (RSS) model is open and technology-neutral. It defines what it means for an autonomous vehicle to truly drive safely by developing a series of logically feasible rules and regulations on appropriate responses to dangerous situations. RSS subverts people's cognition, transforming their inherent understanding of safe driving into a transparent and verifiable mathematical formula, and will redefine the safety standards for future autonomous vehicles.

The decision-making ability of self-driving cars was previously difficult to discover. This ability is mainly driven by a series of artificial intelligence algorithms (such as "black boxes") and is the core intellectual property of leading companies in the autonomous driving industry. The "black box" nature of autonomous driving policies makes it almost impossible to establish a unified autonomous driving safety standard. As some industry experts have said, after cars enter unprecedented development scenarios, statistical bases such as mileage, frequency of manual intervention or simulation hours are no longer sufficient. The introduction of IEEE standards will solve these problems, truly standardize and guarantee the safety of autonomous driving, and ensure people's travel safety.