Framework for Assessing Potential Safety Impacts of Automated Driving Systems (2022)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

76 This report provides practitioners (e.g., transportation infrastructure owners, safety agencies, and ADS manufacturers) with a safety assessment framework to use in safety planning, design, operational, and investment decisions on multimodal infrastructure. The framework accounts for multiple timelines and is developed so that it can be easily applied by practitioners through a series of successive steps. The first step is identifying the ADS feature that agencies would like to assess. The second step is to understand different aspects of the ADS feature such as functionality and expected market penetration rates. The third step is to envision the possible deployment scenarios for the feature. This step builds on data and information shared by car manufacturers and technology companies regarding the operational design domain (ODD) of the feature and the possible timelines of deployment. As part of defining the deployment scenarios, it is important to understand the technology and its infrastructure dependencies, which helps to recognize the associated risks and opportunities. With the scenarios identified, the fourth step is to define safety goals and the associated hypothesis. The fifth step involves the identification of data sources and metrics to help evaluate the stated hypothesis. The analysis method is also chosen during this step to derive insights from the data. Upon completion of the analysis, the results are interpreted to under- stand the safety impact of the ADS feature and to inform related decisions. The report shows the practical application of the framework by summarizing the results of a proof-of-concept study that involved piloting the framework in partnership with two state departments of transportation. One study evaluated a low-speed shuttle scenario in Minnesota, and the other evaluated the potential deployment of ADS-equipped trucks in Virginia’s I-81 corridor. Appendix B demonstrates how the framework could be applied for three additional ADS applications, namely, conditional traffic jam assist, highway truck platooning, and fleet-operated automated driverless vehicles. The framework is applied for two different planning horizons: the short term (0–5 years) and the medium term (5–10 years). It is worth mentioning that four of the five features discussed are envisioned to follow a high-disruption scenario and the numbers in Chapter 4 reflect this assumption. Though this seems to be a reasonable assumption for some jurisdictions, it may not be the same for others. ADS adoption will likely be uneven, both geographically and temporally, and different jurisdictions may experience different dis- ruption levels. Therefore, it is recommended that practitioners review their jurisdictional and institutional goals and policies to identify the possible level of disruption based on the factors mentioned in Chapter 3 (customer acceptance, willingness to pay, policy and regulation, and willingness to share). The infrastructure owners and operators should be aware that their infrastructure condition and investment plans are a key factor for expanding the ODD defined by vehicle manufacturers C H A P T E R 5 Conclusion

Conclusion 77   in the short term. For instance, investing in the infrastructure to maximize the portions of the network that are with good pavement, marking, and signage conditions would maximize the ODD of the fleet-operated automated driverless vehicles in the near term, which would trans- late into saving more lives in the near term. However, as they invest in the key infrastructure elements to enable a particular scenario or deployment level, they need to be aware that car manufacturers are regularly upgrading their technologies. The manufacturer’s goal is providing a driverless vehicle that is capable of navigating the roads with minimal reliance on upgrades to infrastructure, typically, capable of navigating roads with adequate infrastructure (roads that can be navigated by human drivers). State and local agencies should be aware that ADS-equipped vehicles, particularly the Level 4 and Level 5, generate massive amounts of high-resolution data that can be applied in various aspects of the vehicle’s operation (safety research, safety assessment, etc.). Collaboration and data sharing between different car manufacturers could provide better opportunities for enhancing safety. However, driverless car manufacturers or startups are not willing to share data nor col- laborate, mainly because of the proprietary nature of ADS development and relevant intellectual property. There is also little data sharing between developers and regulators or researchers. State and local agencies looking to accommodate ADS features on their roads should con- sider the landscape of the driverless vehicles industry. They will have to consider what level of information to be mandated for disclosure or sharing by the manufacturers and what are the possible ways to audit the reported data without compromising the intellectual property of the manufacturer. For example, California, by law, mandates all companies that are actively testing Level 4 driverless vehicles on California public roads to disclose the number of miles driven and the frequency in which human safety drivers were forced to take control of their autonomous vehicles (Autonomous Vehicle Disengagement Reports). Manufacturers are also required to provide the California Department of Motor Vehicles with Traffic Collision Involving an Auto- nomous Vehicle (form OL 316) within 10 days after the collision (Autonomous Vehicle Col- lision Reports). Once these driverless vehicles are commercially deployed at wide scale, these reports could provide valuable data for refining the framework results. The framework presented in this report was developed considering three aspects: usability (e.g., is the framework intuitive and approachable by the target audience), practicality (e.g., does the framework allow agencies to assess current questions), and data availability (e.g., do current data capabilities limit the application of the framework). This framework was refined throughout the course of this project based on feedback from stakeholder engagement (obtained in research Phase 3) and the proof-of-concept pilots (obtained in research Phase 4).