This project identified the following research needs and suggestions from stakeholder interviews concerning uncertainties regarding AV deployment and the potential mix with human-driven vehicles.

1. Tighten Traffic Control Devices (TCD) uniformity for AVs and develop Manual on Uniform Traffic Control Devices (MUTCD) material to support AV deployment. There’s a need to evaluate how standards such as MUTCD might evolve to meet the development needs of AVs and to allow for more rapid updates to the MUTCD than the current method allows. Tightening uniformity in these areas will help provide more robust marking detection and fewer false positives and will represent an initial step in preparing roadways for AV technologies.
2. Attain machine-vision standards for TCDs regarding size, color, daytime appearance, and nighttime appearance to help support sensor recognition for current Advanced Driver Assistance Systems (ADAS) and future Automated Driving Systems (ADS).
3. Update and maintain pavement markings for today’s ADAS as well as for the future. It is important to quantify the return-on-investment of making national changes (in design and/or maintenance) to any physical infrastructure element but particularly pavement markings.
4. Apply a safe-systems approach to AV deployment for roadway safety since there will be a mixed fleet of human-driven vehicles, including those with ADAS and ADSs using the roadways for decades.
5. Establish AV test scenarios with representative infrastructure conditions. Current testing is conducted under ideal conditions with uniform and dry pavement and clear and dry weather conditions, while in reality, testing conditions for AV technologies should be representative of the existing roadway network that those AV technologies are expected to be used (including adverse conditions).
6. Assess AV readiness by developing a national traffic-control inventory and condition-assessment protocol.
7. Develop Road safety audits (RSAs) to identify opportunities for improvements in safety for all road users. It is important to consider how AVs differ from human-led vehicles and how to accommodate those needs in the training and support material for RSAs.