Implications of Connected and Automated Driving Systems, Vol. 2: State Legal and Regulatory Audit (2018)

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.

NCHRP 20-102(07) Interim Report 10 Method We first discuss our most important methodological choices and then describe the specific methods used for this audit. 3.1 The Research Question The focus of Task 3 was on identifying legal provisions in state codes that might become, or that are becoming, obsolete, and may therefore require modification as C/ADSs become more commonplace. This seemingly straightforward focus proved to be complex in execution. The initial formulation of the task—to identify state laws in need of modification—was understood to be a focus on state laws that might impede the use of C/ADSs to the detriment of broader public benefits. For example, a state legal regime might ban the use of driving automation systems even after the technology is proven to be safer than a human driver. Framed this way, Task 3 was expected to explore state codes with one question in mind: Do the state legislative requirements serve as a barrier to the use of C/ADSs on public roadways? However, in the course of reading state codes, it became clear that the task was much more open- ended. An overly narrow focus might neglect some of the most important laws in need of modification. For example, it is important for state policy makers to know if their motor vehicle codes exempt C/ADSs from legal requirements that apply to conventional vehicles. Similarly, while existing laws may not impede the use of C/ADSs, they may deprive potential beneficiaries of this technology (e.g., those not able to drive currently for medical or other reasons). There are also scenarios where the law may increase the likelihood of C/ADS-related accidents or disputes over responsibility. For example, yielding to blind pedestrians is a common requirement in state codes, yet it is possible that they will not be identified or afforded special attention by C/ADSs. While rules of the road and related requirements may not serve as barriers to the use of C/ADSs per se, some existing laws may complicate the successful deployment of C/ADSs on the roads. As a result, we ultimately included a broader set of reasons for provision modification in our research. Given that the goal of the study was to provide information for DOTs, DMVs and law enforcement administrators, our goal was to identify as many issues and problems as possible. As a result, we erred on the side of false positives over false negatives, opting to identify too much rather than too little. The broader view of the study is illustrated below in Table 2.

NCHRP 20-102(07) Interim Report 11 Table 2. Broader View of Task 3 Adopted in this Study Does the Law: Yes No Serve as a barrier to the use of C/ADSs on state highways (at any or all levels)? Fail to treat C/ADSs symmetrically with human-driven vehicles (e.g., regarding rules of the road)? Fail to enable some of the uses of new technology, negating potential benefits? Place superfluous requirements on C/ADSs that do not impede but are unnecessary or excessive (e.g., mandating that steering wheels meet legislated specifications)? 3.2 General Methodological Choices Fifteen States As discussed in more detail below, we considered the laws and regulations in only 15 states. These states were selected to offer a range of legislative activity throughout the U.S. However, during the crafting of this report, other states not on this list were beginning to introduce legislation (some of which has yet to pass out of the state legislature or awaits executive signature) governing the integration of C/ADSs onto state roadways. Some states within our sample are also in the process of developing regulations to address emerging C/ADS technology. For example, California has proposed regulations for deployment, although at the time of this writing these regulations have not been finalized. In addition, federal agencies are continuing to develop strategies for dealing with C/ADS technology and the U.S. Congress has begun to draft bills governing C/ADSs. Accordingly, analysis in this area cannot be fully contemporaneous. Moreover, important legal challenges that fall outside of our sample and timeframe could be missed and may only be identified after more states are examined. DMV and DOT Focus As many readers are aware, a plethora of legal issues are arising at the intersection of C/ADSs and the law. Issues of federal preemption and other existing federal standards for vehicles and traffic control devices, criminal acts that utilize a car, and even the constitutionality of searches of owners and their vehicles, are all cutting-edge issues emerging in C/ADS research. However, we were not tasked with addressing these or a host of other important questions. Rather, our investigation explores only a narrow slice of the intersection between the law and C/ADSs—the Motor Vehicle Codes (MVCs) and the legal responsibilities of DMVs and DOTs that might be impacted with respect to licensing and titling of the vehicles, licensing the “driver,” reviewing cybersecurity issues, and plans to accommodate C/ADSs within the transportation network. Readers are forewarned, however, that some of these state issues intersect with existing federal standards and developments, particularly with regard to vehicle standards, commercial vehicle restrictions, and traffic control devices.

NCHRP 20-102(07) Interim Report 12 Assumption on Preemption Currently a delegation of federal and state government authorities provides oversight of the management, licensing, titling, and authority to manage automobiles and drivers. The Research Team took the position that NHTSA’s September 2016 Federal Automated Vehicles Policy delegation of duties and authorities between the federal and state governments will not be changed.12 NHTSA notes in the policy that under current law, manufacturers bear the responsibility to self-certify that vehicles they manufacture for use on public roadways comply with Federal Motor Vehicle Safety Standards (FMVSS). If a vehicle is compliant within this framework and maintains a conventional vehicle design, there is currently no specific federal legal barrier to level 4–5 ADS-equipped vehicles being offered for sale. NHTSA’s model state policy confirms that states retain their traditional responsibilities for driver licensing, vehicle licensing and registration, traffic laws and enforcement, and motor vehicle insurance and liability regimes. Modification Rather than Amendment While there is a fine line between the need for a wholly new amendment to existing law as opposed to a modification of an existing provision, we do not consider in this report the need for entirely new laws (e.g., regulating cybersecurity) that may be needed to address the novel risks posed by C/ADSs. These new challenges fall beyond an analysis of existing code provisions since there is no consistent benchmark or trigger for identifying when a new program might be useful. Instead, the need for a new program or requirement hinges on a number of policy and technical factors that fall well outside the scope of this audit. 3.3 Legislative and Regulatory Inventory Audit Method In this task, we reviewed the statutes and regulations in 15 states as well as the UVC for provisions that may need to be modified if states are to provide comprehensive legal oversight for the deployment of C/ADSs on public roadways. Each step is discussed in more detail below. Step 1: Selection of States The first step was to select a range of states that provided a more representative view of the states throughout the country. Since our focus here is on existing state codes that may need modification, we consider not only those states at the forefront of legislative or regulatory activity, but also those that have been relatively inactive. Five states had developed relatively elaborate legislative action as of November 1, 2016 (“Category 1 states” – high levels of legislative activity). Five states generally lacked elaborate legislation as of November 1, 2016, but were actively engaged in testing and/or deployment of C/ADSs (“Category 2 states” – low 12 National Highway Traffic Safety Administration (NHTSA). (2016). Federal Automated Vehicles Policy: Accelerating the Next Revolution in Roadway Safety. Washington, DC: National Highway Traffic Safety Administration, available at: https://www.transportation.gov/AV/federal-automated-vehicles-policy- september-2016

NCHRP 20-102(07) Interim Report 13 levels of legislative activity). The final five states appear to fall at the opposite end of the spectrum in terms of their legal activity and general engagement with C/ADSs (“Category 3 states” – minimal to no legislative activity). The first 10 of these 15 states were targeted as part of the Task 2 stakeholder outreach efforts.13 We also included the latest edition (2000) of the entire UVC in our audit. While adoption of the UVC by the states varies widely depending on the provision (some provisions are almost uniformly adopted; others are almost uniformly rejected), the UVC provides a good starting point for examining state codes more generally. Particularly given our limited sample of 15 states, a review of the UVC may capture approaches taken by states not in our sample. For that reason, in our findings and analysis section, we typically open our discussion of each area of the law with the UVC approach.14 To select the Category 3 states (minimal to no legislative activity) we applied the following general criteria for inclusion: 1. No significant (i.e., minimal to no) legislative action had occurred in the state with respect to automated transportation as of November 2016. 2. The state fell at the lower range (25th percentile) of either population size or median household income or both. The state also fell in the lower third of state gross domestic product in 2016.15 3. The state fell in the lower third of population density (low density).16 Both criteria 2 and 3 make it more likely that the challenges posed by C/ADSs will be relatively low on the state’s list of legislative priorities. From the set of states meeting these criteria, we ultimately selected five states for inclusion in the 15-state sample. Our ultimate selection of the five states was not purely random; instead, we 13 California, Florida, Michigan, Nevada, Ohio, New York, Pennsylvania, Texas, Utah, Virginia 14 Using the UVC is not an endorsement of its potential use in states. Additionally, the UVC is clearly outdated, with its latest revision date occurring nearly two decades ago. Although it is outside the scope of our audit, readers should be aware that the Uniform Law Commission (ULC) will be drafting a set of “model uniform laws” for level 4–5 ADS-equipped vehicles within the next 2 years. See, e.g., Uniform Law Commission, New Drafting and Study Committees to be Appointed (July 30, 2014), available at http://www.uniformlaws.org/NewsDetail.aspx?title=New%20ULC%20Study%20and%20Drafting%20Committees %20will%20be%20Appointed. Other organizations such as the American Association of Motor Vehicle Administrators (AAMVA) and NHTSA (via the Manual on Uniform Traffic Control Devices for Streets and Highways [MUTCD]), are also actively involved in considering best practice guidelines and other sources of assistance to help guide states in the near future. See, e.g., MUTCD, available at https://mutcd.fhwa.dot.gov/kno_2009r1r2.htm and AAMVA Working Group, available at http://www.aamva.org/Autonomous-Vehicle-Best-Practices-Working-Group/. 15 We used U.S. Census data for median income (https://www.census.gov/data/tables/time-series/demo/income- poverty/historical-income-households.html) and U.S. Department of Commerce data for gross domestic product (http://bit.ly/2vXGbxz). 16 We used the rankings provided here: http://state.1keydata.com/state-population-density.php

NCHRP 20-102(07) Interim Report 14 attempted to include a mix of states that vary geographically and by terrain.17 (In the course of our analysis we discovered that this was a good decision, as rules of the road vary among states with respect to conventions for farm vehicles, livestock, mountain curves, etc.) Our final list of Category 3 states consists of states in the southwest (New Mexico); north (Wyoming); Midwest (Nebraska and South Dakota); and south (Mississippi). Wyoming was added not only because of its low population density and varying terrain, but also because of its relatively high per capita income. Our total list of Category 1, 2, and 3 states is provided in the following table (Table 3). Table 3. States Included in the Legal and Regulatory Audit State Category Description States Included Category 1 States with targeted legislation that clarifies the term “operate” or “drive” in place as of November 1, 2016, for all levels of C/ADSs and both testing and deployment. California Florida Michigan Nevada Ohio Category 2 States generally lacking elaborate legislation as of November 1, 2016, but in which pilot tests have occurred, and in which executive orders have been issued requiring studies to be undertaken. New York Pennsylvania Texas Utah Virginia Category 3 States with minimal to no legislative or executive activity, but where studies may have been undertaken by the DOT. Mississippi Nebraska New Mexico South Dakota Wyoming The resulting 15 states in this audit offer a range of legislative approaches to C/ADSs. The Category 3 states at one end of the spectrum operate under statutes and regulations that were passed with presumably no anticipation of C/ADS challenges. In the middle of the spectrum are Category 2 states like Texas that have been actively involved in testing C/ADSs, but have not passed significant legislation as of November 2016. These states operate under state codes that are not much different from those in the Category 3 states; however, codes in Category 2 states have included some provisions for C/ADSs. At the far opposite end of the range are Category 1 states like California, Michigan, Nevada, Ohio, and Florida, which have engaged in considerable legislative activity to accommodate C/ADS technologies. For purposes of spotlighting laws that 17 We resisted a pure “random” selection of the five states remaining in large part because we were concerned it would produce states that were too geographically similar – like South and North Dakota or Alabama and Mississippi. Hand selecting the final candidates allowed us an opportunity to gain more variation in the final five in terms of general differences.

NCHRP 20-102(07) Interim Report 15 are “in need of modification,” these Category 1 states tend to offer a useful point of contrast for how codes might be changed in the future. Step 2: Locating Provisions in Need of Modification Our initial and most substantial work was dedicated to a thorough read of 15 state codes. We started with the “laws,” identifying provisions that might need to be modified because they impeded automated transportation functionality or they were, or would be, no longer relevant. We conducted the audit of each state code by reading through each state code from front to back. Each code-reader employed a qualitative list of triggers (provided in Appendix 2) to help identify these provisions. Please note, that while systematic by its nature, the work was ultimately subjective.18 Initially, we created our inventory of legal provisions analyzed in Chapter 4 by extracting provisions that could be problems in the future and would therefore require modification. Trained law student research assistants conducted this work for four of the 15 states. The lead author, Wendy Wagner, and a licensed lawyer, Susanna Gallun, conducted the coding work on the remaining 11 states. As work progressed, we also created a more comprehensive Microsoft Word file that included not only the provisions that we singled out, but also those that we did not believe needed modification. Word files of all of the provisions that underwent this closer scrutiny, along with comment boxes and notes are available online.19 Wagner and Gallun also relied on the trigger list (Appendix 2) to help identify provisions in need of further study. Once all the code provisions that appeared in need of modification were identified, Wagner sorted the provisions into general categories and subcategories. The findings and analysis in Chapter 4 are based on these extracted groupings of provisions. After the analysis of state laws (codes) was completed, we conducted a review of state regulations. We first conducted a review of legislative programs in search of specific regulations that might a) require modification to avoid hindering the advancement/deployment of level 4–5 ADS-equipped vehicles, b) require significant amendments to meet existing safety goals, or c) might become irrelevant as fleets utilizing level 4–5 ADS-equipped vehicles grow in size. These programs included vehicle registration, driver testing, driver education, vehicle inspection, reciprocity requirements, and aftermarket modification. We then searched the remaining state motor vehicle regulations for general provisions that also might require modifications for the same reasons, or might create unreasonable civil or criminal offenses for an occupant or user of a level 4–5 ADS-equipped vehicle while it is appropriately deployed in automated mode. 18 Since this project attempts to chart the legal landscape with regard to state laws in need of modification in the wake of automated transit, the assurance of precision in coding from state to state seemed much less essential. We thus did not test for inter-coder reliability. 19 Available at: https://utexas.box.com/s/341xa53e7yb8usyv0vjkjhtc45mtqfdr

NCHRP 20-102(07) Interim Report 16 Step 3: Expert Ground-truthing and Peer Review The preliminary findings and analysis from Step 2 were shared with trusted regulators and legislators, stakeholders, and academic experts to ensure that the analysis was complete and to obtain input/feedback. This process included a draft review by the Virginia DMV from the perspective of a DMV relying on this research to assist in implementing an overall strategic direction for automated transport. We selected the Virginia DMV for this more comprehensive review because Virginia was among the 15 states studied and the Virginia DMV indicated a willingness to invest the time of several staff members in the effort of providing a multi-person, front-to-back review of the draft report. The Virginia DMV was also willing to provide the review in a short time period (roughly one month). Limitations Note that this general method, which involved an extensive reading of state codes in search of problematic provisions, is subject to several limitations. Even after culling hundreds of extraneous sections within a single state code, each state code still contained several hundred provisions requiring close review.20 (We estimate that well over 5,000 separate legislative provisions were ultimately examined in the course of this audit). Human error when reviewing so many individual sections is inevitable. Additionally, as lawyers are well aware, a comprehensive review of a state legal program will miss some important nuances. Legislatures sometimes place legal exceptions in strange places (and in some cases this legislative drafting technique is not an accident). Such exceptions might wholly change the meaning of other provisions in ways that can be discovered only after a careful parsing of the text. While these limitations are unavoidable in an audit such as this, they are important to keep in mind in interpreting the findings. To assess the status of any individual state, including those in our sample, one would need to conduct a much more thorough and court-grounded reading of individual state codes. We also hasten to add that our search was exclusively for provisions in need of modification. If a state had already amended (i.e., modified) its law specifically to address legal challenges posed by C/ADSs, then these provisions were not included in our search. We have, however, used some of these new provisions in the course of the discussion to offer a point of contrast to existing approaches that need modification. As just one example, the majority of states have what are known as “visual display” restrictions on the use of television or video monitors. These monitors cannot be visible to the driver. Virginia recently amended its law to allow visual displays for the driver while a level 3–5 ADS-equipped vehicle’s system is properly engaged.21 Virginia’s current law does not, therefore, require modification, but is illustrative of one viable legal approach that other Category 2 and 3 state legislators and agencies might look to. 20 Wagner read the section titles of all provisions within each state motor vehicle code and made an initial decision about those sections to exclude from the downloading process. The files uploaded at the previous footnote contain only those sections Wagner believed were potentially pertinent, while still erring on the side of over-inclusion. Thus, only about one-half to two-thirds of the entire motor vehicle codes were ultimately downloaded. 21 See, e.g., VA HB 454, available at http://lis.virginia.gov/cgi- bin/legp604.exe?ses=161&typ=bil&val=hb454&submit=GO