National Highway Traffic Safety Administration
1200 New Jersey Avenue, S.E.
Washington, DC 20590–0001
RE: Automated Driving Systems: A Vision for Safety
Docket No. NHTSA–2017–0082
Dear Mr. Wood,
On behalf of the Transportation Trades Department, AFL-CIO (TTD), I am pleased to provide comments on the National Highway Traffic Safety Administration’s (NHTSA) A Vision for Safety 2.0 guidance regarding autonomous vehicles. By way of background, TTD consists of 32 affiliate unions representing workers in all modes of transportation, including those who will be impacted by the development and deployment of automated driving systems. We therefore have a vested interested in the guidance. 
NHTSA states that the purpose of the guidance is to support industry innovators, States and other key stakeholders as they consider and design best practices relative to the testing and deployment of automated vehicle technologies, as well as to update and refine the 2016 Federal Automated Vehicles Policy. TTD understands that NHTSA’s guidance is both voluntary and will continue to be updated going forward. However, we believe that the guidance as published is lacking several components that should be included to ensure the safe performance of any automated driving system (ADS).
In A Vision for Safety 2.0, NHTSA correctly identifies human-machine interface (HMI) as an important component of the design process. To this end, NHTSA recommends that ADS systems have the ability to inform a driver of its operability, when it is in use, and when the system requests the driver take back control. The agency also suggests manufacturers consider if driver engagement monitoring is appropriate. While these components are necessary, they are not sufficient to maintain human operator engagement. NHTSA should clearly encourage manufacturers to engineer HMI to ensure that drivers are engaged to the standard necessary for safe operation. Most notably, ensuring a safe transition between machine and human operator should be of paramount concern for the agency and the manufacturer, especially given recent crash incidents, including the Tesla accident of May 2016. Last second warning systems, which may be too late, and passive solutions like owner’s manual disclaimers are not sufficient on their own to mitigate the distraction and driver disengagement risks that come with ADS technology. NHTSA’s guidance should recommend manufacturers design HMIs engineered to overcome these issues.
In updating its guidance, NHTSA has removed references to data sharing and publishing that were included in the 2016 document. While TTD appreciates that manufacturers have concerns about divulging trade secrets in a competitive industry, we firmly believe that NHTSA should amend its guidance to encourage the availability of certain data, particularly post-crash and near-miss information to best inform the public, legislators, and regulators. Much as current crash databases serve to inform regulators on vehicle standards, post-crash and near-miss information for ADS offers regulators the 21st century version of crash test dummies. The specter of proprietary information should not stand in the way of NHTSA collecting the information it requires to fully understand any risks or shortcomings of ADS technology and improve system safety. We encourage the agency to promote the collection and sharing of data through this guidance.
As ADS technology moves forward, design standards must ensure that a driver can operate the vehicle safely when needed, and that any automated technology improves, rather than diminishes safety. By offering stronger recommendations on HMI design and its function, as well as on data sharing requirements, NHTSA can better promote the development of autonomous vehicles that put safety first. We thank NHTSA for the opportunity to comment on this guidance and look forward to working with the agency and DOT’s other modal agencies on autonomous vehicle issues going forward.
 Attached is a complete list of TTD’s 32 affiliate unions.
TTD Comments on NHTSA Autonomous Vehicle Guidance (85kb)