Analysis of the USDOT’s Regulatory Review for Self-Driving Cars (Part 1): References to “drivers” in the federal regulations

Editor’s Note: Apologies for the unannounced gap between posts.  I have been on parental leave for the past two weeks bonding with my newborn daughter.  In lieu of the traditional cartoon, I will be spamming you today with a photo of Julia (see bottom of post).  Now, back to AI.

The U.S. Department of Transportation recently released a report “identifying potential barriers and challenges for the certification of automated vehicles” under the current Federal Motor Vehicle Safety Standards (FMVSS).  Identifying such barriers is essential to the development and deployment of autonomous vehicles because the manufacturer of a new motor vehicle must certify that it complies with the FMVSS.

The FMVSS require American cars and trucks to include numerous operational and safety features, ranging from brake pedals to warning lights to airbags.  It also specifies test procedures designed to assess new vehicles’ safety and whether they comply with the FMVSS.

The new USDOT report consists of two components: (1) a review of the FMVSS “to identify which standards include an implicit or explicit reference to a human driver,” which the report’s authors call a driver reference scan; and (2) a review that evaluates the FMVSS against “13 different automated vehicle concepts, ranging from limited levels of automation . . . to highly automated, driverless concepts with innovative vehicle designs,” termed an automated vehicle concepts scan.  This post will address the driver reference scan, which dovetails nicely from my previous post on automated vehicles.

As noted in that post, the FMVSS defines a “driver” as “the occupant of a motor vehicle seated immediately behind the steering control system.”  It is clear both from this definition and from other regulations that “driver” thus refers to a human driver.  (And again, as explained in my previous post, the NHTSA’s recent letter to Google did not change this regulation or redefine “driver” under the FMVSS, media reports to the contrary notwithstanding.)  Any FMVSS reference to a “driver” thus presents a regulatory compliance challenge for makers of truly self-driving cars, since such vehicles may not have a human driver–or, in some cases, even a human occupant.

The driver reference scan portion of the USDOT’s review was designed to identify every instance in the FMVSS where the regulations refer to a driver, whether (a) explicitly by including the word “driver” or the equivalent term “operator,” or (b) implicitly by including language that clearly assumes the presence of a human driver.  An example of the latter is the FMVSS requirement that service brakes must “be activated by means of a foot control”–no one would seriously dispute that the “foot” in question is that of a human driver, even though the word “driver” never appears in the rule’s text.

Summarizing the findings from the driver reference scan, the report’s authors noted that “as long as [a] vehicle design allows a human driver to operate the vehicle with a wheel and pedals,” most of the FMVSS’ driver references “do not preclude certifying a vehicle with automated capabilities.”  Notably, however, the Google Car does not have a steering wheel or accelerator/brake pedals, instead giving the vehicle’s AI system exclusive control over vehicle operations.  Such advanced automated vehicle designs “may face significant challenges to certification under the existing standards,” according to the report’s authors.

The authors also warned that the FMVSS’ current definition of the term “driver” could create interpretation issues, since the definition is premised on a human driver in the front-left seat having steering control:

[T]he definition of the term “driver” would require further interpretation in the context of a vehicle that is occasionally or exclusively controlled by advanced software. Currently, the definition of a “driver” means the occupant of the motor vehicle seated immediately behind the steering control system” – §571.3), but how this definition is interpreted could become less clear as highly automated vehicles takeover more of the driving function.

In total, the driver reference scan revealed that 33 of the 73 standards in the FMVSS–including 19 of the 29 standards pertaining to crash avoidance–either explicitly refer to a human driver or implicitly assume that the human driver is present and in control of the vehicle.  But this does not mean that there are only 33 references in the FMVSS to human drivers.  On the contrary, because each standard in the FMVSS consists of multiple subparts, many standards contain numerous driver references.  Based on my count, the report identified a total of 286 references to human drivers in the FMVSS.

The report identified a number of different types of driver references, including:

1. Requirements that a piece of vehicle equipment must be visible to the driver or must communicate information to the driver in a specific way (e.g., an audible-to-the-driver warning must sound when the driver’s side door is opened while the key is still in the ignition).  The authors identified 72 such references in the FMVSS, including 17 in Standard 101 (which includes rules for hand-operated controls and for displays such as warning lights) and 14 in Standard 138 (tire pressure monitoring).
2. Requirements that a control mechanism must be operable by the driver (e.g., the foot-control brake requirement).  This was the most common type of driver reference, with 93 references of this type listed in the report.  There are especially high concentrations of driver operability references in standards 122 and 135, which respectively cover braking systems for motorcycles and light vehicles (i.e., cars and trucks).
3. Requirements that the driver must be able to observe the outside environment using vehicle equipment (e.g., rules governing glazing materials used on car windshields).  The report identified 32 instances of this type of driver reference; unsurprisingly, most (21) appear in Standard 111, which covers rear visibility.
4. Requirements referencing the driver’s physical state or position within the vehicle.  Examples of this include the requirement that dashboard warning lights must be visible to a driver who is properly buckled into his seat, and various test procedures requiring that the “driver’s seat” be placed in a certain position prior to testing.  There are 40 references of this type in the FMVSS, including 11 in Standard 111 (rear visibility) and 15 in Standard 208 (crash protection, which includes rules for seat belts and airbags).
5. NHTSA safety testing procedures that reference a necessary action by or capability of the human driver.  The FMVSS include 49 such references, with the vast majority appearing in three standards relating to brakes (105, 122, and 135), often in the context of specifying the amount of force that should be applied to a brake pedal during testing.

The first striking aspect of the driver reference review is the sheer number of driver references that appear in the FMVSS.  This reflects that the presence of a human driver is not merely an assumption that underlies the FMVSS, but instead is a necessary condition for compliance with many of the FMVSS’ provisions.  Eliminating the necessity of a human driver from the FMVSS will require literally hundreds of amendments to the FMVSS’ regulations–amendments that can only be made through the laborious, time-consuming federal rulemaking process.

Moreover, most of the necessary rule changes could not be achieved simply by redefining “driver” for purposes of the FMVSS.  Because an AI system would not use a steering wheel, brake pedals, or a gear stick, many NHTSA standards regarding driver inputs and test procedures would be impossible for truly self-driving cars to replicate (at least without equipping self-driving cars with unnecessary robotic arms and legs).  The easiest way to do this would be to simplify the test procedures to focus simply on the end result that must be achieved (e.g., specifying the amount of time that it takes for the vehicle to come to a stop after activation of the brakes) without regard to any specific physical inputs (such as the amount of force applied to a brake pedal in order to bring the vehicle to a stop).  But I am less familiar with FMVSS’ test procedures than I am with the FMVSS’ equipment and safety feature requirements, so it is difficult for me to say how difficult it would be to come up with new test procedures that could be applied to autonomous vehicles.

Stay tuned next week for the second part of this analysis, focusing on the USDOT report’s review of various proposed concepts for autonomous vehicles and the regulatory challenges each concept might face.