Note: The National Highway Traffic Safety Administration (NHTSA) and the Department of Transportation issued an Advance Notice last November on a project to standardize the safety of self-driving cars. We asked Egil Juliussen, a seasoned automotive analyst and EE Times columnist, to tell us about this. In this text, he will explain to us how the NHTSA defines self-driving cars, what constitutes “safety standards,” and what questions the government is asking the industry to improve its bill.
On November 19, 2020, NHTSA published a preliminary version of its bill with safety standards for self-driving vehicles... Comments and amendments to this document are accepted until February 1, 2021. Work on this document marks the beginning of the standardization of rules for the operation of self-driving cars. The recommendations in this document may become some of the most important for the automotive industry for years to come. What is NHTSA striving for and what can we expect from this project?
The title of the bill sounds like "Safety Fundamentals for Ride Automation Systems." The official version of the document , more than 60 pages in length, was included in the Federal Register on December 3, 2020 (18-page section in small print). There is also a link where you can read public comments, and you can leave your own here .
The management uses the term Automated Driving System (ADS) to describe the hardware and software of unmanned vehicles. Further in the text I will also stick to this term.
Summary of the document
The next paragraph is a summary of the ADS bill.
(ADS). , ADS, . , ADS . , , .
Written comments must be sent by February 1, 2021 to the Federal portal for electronic lawmaking: http://www.regulations.gov . Comments should reference case number NHTSA-2020-0106.
Prerequisites for the creation of a bill
The Office came to work on this document after collecting large amounts of data and statistics in the field of unmanned vehicles. The first federal vehicle safety standards were adopted back in 1966, and even then, the department had statistics on road accidents that allowed it to determine regulations to reduce the number of accidents. Today, there is much more data on accidents, but there is not enough data to develop driver assistance systems and other safety systems.
Developing standards for ADS is more difficult for a number of reasons:
- There are no vehicles with ADS that the management could test and draw conclusions
- There is no operational data for management to analyze
- Ride automation technology is advancing rapidly, which could make safety standards obsolete in 10 years.
- The department lacks technical expertise to develop its own ADS
Challenges facing NHTSA
The NHTSA, as they say, is caught between a rock and a hard place. There is already criticism in the public field, which says that the document published by the administration is too weak, loyal to the automotive industry, and does not contain technical details and safety rules.
Some of these comments are valid, but they do not take into account the constraints the department had to deal with — budget constraints, lack of technical resources, and political issues.
Management needs to enlist the support of Ride Automation companies because they have the experience and technical expertise and because this market will continue to grow. The NHTSA will be on the right track by working with the developers of ADS to develop security standards for ADS.
A governance approach to standards development
The NHTSA has researched self-driving technology and has published several publications on the topic. Most of these publications are devoted to what requirements are related to the safety of vehicles with ADS. The department is also finalizing and adapting testing procedures for vehicles with ADS.
How management defines ADS
ADS is hardware and software that together can perform the task of dynamic driving, regardless of the constraints of the vehicle's operating areas.
In less technical terms, ADS provides driving and control functions for the situations for which the system is designed.
What are the security standards for ADS?
The NHTSA bill is a complex 64-page document with regulatory restrictions, technical details, an analysis of what didn't work, a discussion of the ADS structure and security rules that will be implemented in the future. The Office closely monitors the impact of technological innovation and strives not to hinder them.
The figure below provides an overview of what is included in the NHTSA ADS security standards. The Office relies on the research and expertise of ADS (automotive and high-tech) companies. The left side shows the actions taken by the management (in black), and on the right - the contribution of companies from the industry (in blue). The red block in the middle is a potential ADS security project.
The blocks in the above figure correspond to the main content of the document. The bill can be divided into two main segments: the technologies used in ADS, and the administrative measures for the implementation and control of these systems. These segments have subsections shown in the figure:
- Technical Management Studies: Research Findings and Reports
- Industry Tech Research: An Overview of Key Self-Driving Technologies
- Industry Technical Studies: Proposed ADS Security Project Structure
- Research of technical processes in the industry (standards)
- Administrative regulatory mechanisms
- Administrative mechanisms: methods of interaction with the industry
The agency wants input from industry and has included 25 questions in the comments. Comments are accepted until February 1, 2021. The perspectives of each of the sections are presented in the figure below.
Technical control measures
The NHTSA has done a lot of research to prepare for the development of ADS security standards. The Office issued four reports:
- AV 1.0: Approaching the Road Safety Revolution (September 2016)
- AV 2.0: Security Concepts (September 2017)
- AV 3.0: Preparing for the Transport of the Future (October 2018)
- AV 4.0: Securing the Leading American Industry in Self-Driving Technology (January 2020)
The agency launched its Self-Driving Vehicle Transparency and Safety Initiative (AV Test) to provide the public with information on ADS development. This website ( AV TEST Initiative | Automated Vehicle Tracking Tool | NHTSA ) is an interactive environment for discovering who is testing unmanned systems and where.
NHTSA launched the AV Test in June 2020 with support from states, local governments and private industry stakeholders. The website provides access to data on state activities, current laws, regulations, measures to promote ADS, as well as data from companies involved in the development and testing of ADS.
The NHTSA study focuses on self-driving safety metrics and aims to identify methods, metrics and tools to measure ADS performance. These estimates include the performance and behavior of systems relative to the declared workspaces and objects in them. It also evaluates recognition capabilities, methods of responding to various events and fault tolerance when leaving the working area.
The Office has developed several ADS security models and metrics. The ISM (Instant Safety Metric) metric evaluates all physically possible trajectories along which a vehicle and other road users can move, taking into account a set of their possible actions. MPrISM (Predictable Instant Security Metric Model) is an updated approach based on ISM that complements the scoring method used in this metric.
Technical Measurement of ADS in Industry: Key Technologies
Vehicles with ADS must perform all the driving functions that drivers perform: “feel” the behavior of other road users, perceive / understand their actions, plan a trajectory, build a path, control movement, speed and estimate the braking distance. Much research in the field of autonomous driving is aimed at finding the right solutions to implement these functions.
ADS Engineering Tests by Industry
The FDA reviewed three industry models for ADS security. Mobileye offered its RSS (Responsible Security) model . RSS is a mathematical model that includes common sense rules for interacting with other road users. The goal of this model is to minimize the likelihood of an accident while operating within normal behavioral expectations.
Nvidia offered another model called the Safety Force Field . This system uses numerical methods to assess situations through simulations - the system evaluates whether the ADS is successfully managing the perception and control of the environment and whether it is not taking inappropriate actions.
In 2018, Rand released a report that proposed security models that looked at ways to define and measure ADS security.
ADS Industry Security Standards
ISO 26262 describes methods for assessing functional safety in the design of electrical and / or electronic systems. It is the main standard for ADAS solutions and future ADS.
The ISO 21448 standard (or "Security of Intended Functions" - SOTIF) is designed to assess functions for which information awareness plays a decisive role (as well as for systems in which information awareness is provided through the use of complex sensors and processing algorithms, especially when it comes to emergency systems). SOTIF does not apply to faults specified in ISO 26262.
UL 4600 Is a technology standard that is intended for use by manufacturers in the development of ADS - it was developed for them in the first place.
: ADS
AV 2.0, published by the agency, introduces the concept of voluntary self-assessment of safety (VSSA). This document identifies 12 safety factors that ADS designers should consider when designing and testing their systems. As of January 14, 2021, 26 companies have submitted their reports (see Voluntary Safety Self-Assessment | NHTSA )
NCAP also plays an important role in safety standardization (especially ADAS). It makes sense to add ADS security assessments in newer versions of the NCAP standards. For example, new versions of their documents could incorporate an obstacle course test - the results will be useful for car buyers participating in the NCAP certification program.
Currently, the technology development prospects for ADS are unclear and depend on many future advances and innovations. Consequently, the corresponding regulatory performance indicators and safety thresholds remain unclear. This is why NHTSA is committed to improving safety through voluntary guidance rather than strict regulations and requirements.
Administrative measures: regulations
The NHTSA has already taken steps to secure data disclosures and communications for exemptions from applicable regulations. An example of such a situation is the fact that Nuro was allowed to deploy ADS vehicles to transport goods.
The agency has broad legal authority in the area of vehicle safety and can use it to “reduce the number of accidents and deaths / injuries resulting from them”. The NHTSA believes that ADS regulation will become necessary at some point and is exploring ways to do it. The Office can create new rules or change existing ones to regulate unmanned vehicles.
Typically, the department used its powers in two ways:
- Or to oblige companies to implement proven technologies in accordance with standards to meet safety requirements, as well as to adjust technologies to minimum requirements
- Or to regulate voluntarily introduced technologies with minimum security requirements
The department is also considering the possibility of introducing a new way of regulating standards in connection with the rapid changes in the field of automotive technology (these changes are associated with the development of automotive software and the introduction of update technology by air). Next-generation road safety standards should provide developers of vehicles, sensors, software and other technologies needed for ADS with sufficient flexibility to change and improve their products without having to constantly adapt regulations.
NHTSA Questions to Industry
Management wants industry companies to be actively involved in the project and invites them to answer 25 questions. The deadline for submitting comments to the published draft is February 1, 2021. As of January 14, 16 comments had been submitted. A comment from the CTA (organizer of CES 2021) requested an extension of the submission deadline until March 3, 2021. The industry needs to do more to share its self-driving experience with management and incorporate it into the new bill!
Security design evolution for ADS
The figure below attempts to describe the future and make predictions about how the NHTSA ADS security standard might evolve. A number of blocks from the previous figure are shown on the left, and forecasts for the future are highlighted in red. There are many questions to this project - they are presented in red and green blocks.
The big question is whether this project will be needed in such ADS applications as autonomous trucks, robo taxis, transportation of goods and, ultimately, ADS for consumers. There is another important question that concerns how this project will develop in accordance with the existing safety standards. I think that these processes will take place in accordance with the picture above. After the presentation of Mobileye at CES 2021 (" Why plans to release a consumer autopilot by 2025 are plausible " ) I added consumer ADSs to the scheme, but only in the context of level 4. Level 5 decisions will appear much later.
Another question concerns the self-government law passed by the House of Representatives (but not the Senate) - which the impact will it have in the future, when / if adopted?
To summarize, I believe the FDA has invested great effort in researching and promoting its safety assessment system in a challenging environment. Industry representatives have to answer 25 questions presented in the document. and apply your experience to improve the bill.
In my next texts, I will analyze the document on safety standards for self-driving vehicles from NHTSA in full detail.
- Russia's first serial control system for a dual-fuel engine with functional separation of controllers
- In a modern car, there are more lines of code than ...
- Free Online Courses in Automotive, Aerospace, Robotics and Engineering (50+)
- McKinsey: rethinking electronics software and architecture in automotive
About ITELMA
List of useful publications on Habré