Traditionally, vehicle operation relies exclusively on a human operator for one or two-way information exchanges. Licensed individuals are trained to become situationally aware and communicate through auditory and visual cues directed at other road users across different modalities of travel (bicycle, foot, train, motorcycle, among others). Drivers gather data through their senses and communicate their intentions to act using the car’s horn, headlights, or turn signals as well as the driver’s voice, posture, facial expressions, or hand gestures.
These behaviors are covered by regulated and nonregulated means. Regulated means include specific rules on the size and light intensity of turn signals. They also cover behaviors required while navigating a vehicle. For instance, the California driver’s handbook includes several mentions of when and how to use turn signals (e.g., when merging lanes, to acknowledge when an officer pulls a person over, and to identify the intended behavior of other vehicles).
Nonregulated means for driver communication exist as social norms that involve the use of gestures and eye contact, among other behaviors, to indicate intent. The exchange of information through social norms provides drivers with significant discretion over how and when to communicate, such as when to honk a horn to gesture approval or disapproval of another person’s behavior.
Increasingly, autonomous vehicles (AVs) challenge the long-standing governance of vehicle navigation. As a term, AV is commonly used as a catchall for vehicles able to complement or replace human operation of a vehicle. The accepted nomenclature to describe this technology’s capabilities was created by the Society of Automotive Engineers (SAE) and features a scale from 0 to 5. Vehicles classified between 0 and 3 depend on individuals to navigate. Meanwhile, AVs at SAE level 4 and above release humans from all responsibilities over AV operations and are approved to commercialize taxi services without operators in a limited number of locations such as Los Angeles, Phoenix, and San Francisco.
Unlike human-driven vehicles, AVs differ in one key respect: those above SAE level 4 do not rely on an individual. While this offers significant opportunities to expand AVs’ communication capabilities beyond those of non-AVs, it also generates risks that should be managed. Regulated and nonregulated exchanges of information must be adapted for a technology that eliminates the human role in vehicle operation, but continues to depend on interaction with humans to complete its objectives. Moreover, sorting AV behaviors into a conceptual framework would contribute in generating normativity in how AV communication is classified by sector stakeholders.
This issue is an example of regulatory novelty for determining the role of governance in replacing humans and managing unknown scenarios of varying complexity. Coping with this technology makes it necessary to identify adaptable and responsive governance mechanisms that can cut the distance between a fast-moving technology and lagging governance, also known as the pacing problem. While enacting hard law is effective in mandating a desired outcome, less so is the significant stakeholder bandwidth needed to create and modify it.
A dynamic alternative to hard law is soft law, which is defined as programs that “create substantive expectations that are not directly enforceable” by government. Soft law mechanisms can flexibly manage growing expectations for AVs. The U.S. government is currently championing a soft over hard law approach through its publication of several AV frameworks.
This Article has two goals. First, it proposes a framework that characterizes the communication capabilities of AVs based on the following dimensions: safety, infrastructure support, information security, and profit motive. Second, it identifies existing opportunities to govern AV communication capabilities through soft law mechanisms. Ideally, this Article will improve how transportation stakeholders in the public and private sector proactively manage this technology. This is critical considering that, in all likelihood, AV capabilities will alter social norms in the same way that the first mechanized vehicles impacted the perception of transportation technologies risks held by individuals accustomed to travel by horse.
I. Characterizing the Communication Capabilities Of AVs
In supplanting humans, AVs labeled as SAE level 4 and above will exchange information with individuals and other vehicles, both AVs and non-AVs. As industrialized countries face a fast-evolving transportation marketplace, decision-makers from within and outside government need access to coherent guidance about which AV communication capabilities demand governance. Support to identify potential management vectors can help stakeholders resolve existing or potential governance gaps in the transportation sector.
Part I proposes a framework that classifies AV communication capabilities and reflects the state of AV soft law governance in industrialized countries, particularly in the United States. It does so by dividing these capabilities into four dimensions: safety, infrastructure support, information security, and profit motive. In addition, it examines the availability of soft law programs to manage each dimension. As a result, it provides stakeholders with relevant information about how much attention has been dedicated to each dimension and determines if more governance is merited. Highlighting the range of possible AV communication activities should empower stakeholders to apply flexible and agile governance mechanisms to resolve governance gaps.
A. Safety
AVs promise a future in which safety does not depend on the capacity, awareness, state of intoxication, or reflexes of an error-prone human driver. In effect, this technology could reduce or eliminate a yearly death toll of almost forty-three thousand individuals due to crashes that are overwhelmingly (93%) caused by human factors. To achieve this safer future, stakeholders must identify how to appropriately govern AVs to maximize their benefits, including those centered on the exchange of information.
Even in complex environments, machines can capture and analyze information at a higher and more consistent rate than humans. Thus, a fundamental component of AV safety is its effectiveness in gathering and communicating information. This Section explores three scenarios where communication is critical to mitigate direct or indirect harm caused by an AV: interactions with pedestrians and individuals in other vehicles, interaction with authorities, and interaction with non-road AVs.
1. Interaction with Pedestrians and Humans in Other Vehicles
Interacting with pedestrians, human drivers, and passengers in motorized or nonmotorized vehicles adds complexity to vehicle operation. The safety of vulnerable populations, such as pedestrians, cannot be underestimated, as they represent over fifty percent of recorded traffic deaths worldwide. In the United States alone, over seven-thousand pedestrians died from traffic incidents in 2022. Pedestrians will be less likely to suffer harm caused by an AV encounter if this technology values pedestrian safety over its navigational objectives. This scenario describes a state of “[p]edestrian [s]upremacy,” where AVs will behave conservatively to avoid harming individuals outside the vehicle.
Optimizing pedestrian supremacy does not account for the full spectrum of safety issues. Humans inside AVs also face risks when an AV favors the well-being of those outside rather than those inside of it. This risk trade-off is why effective communication serves as a vector to minimize risky interactions. In the context of AV interactions with humans in other vehicles, there are two areas where soft law can enhance vehicle safety: distinguishing AVs from other vehicles and communicating AV intent to humans. For both, widely adopted soft law programs have yet to emerge.
First, soft law can enhance vehicle safety by improving how AVs eventually communicate their capabilities. At present, there are no clear indicators for human identification of an AV’s ability to perform at an SAE level 4 or above. Even vehicles performing below an SAE level 4, such as Tesla’s “Autopilot” or General Motor’s “Super Cruise,” lack any method to indicate that these systems are engaged to individuals outside the vehicle. Thus, pedestrians and other drivers are left to assume that a vehicle’s equipment (e.g., radar), branding, or the visible lack of a driver might be a sign of autonomy.
Automakers would benefit from standard external indicators that signal when a human is not in charge of AV decision-making. Doing so would allow people to exercise the necessary caution when interacting with this technology. In this respect, the state of California has clear requirements for recognizing AVs that are being tested. However, these guidelines are limited to information provided to local authorities on the model, make, license plate, and vehicle identification number. Unfortunately, none of these improve the public’s awareness of AVs. Thus, there is an opportunity to create and implement soft law for AV identification standards so human drivers and pedestrians can make informed choices.
Second, soft law can enhance vehicle safety by standardizing how AVs communicate intent. AVs can theoretically process data consistently and continuously, but if their sensors are unable to capture information because they are blocked by incoming traffic, a tree, or other sources of signal pollution, then their ability to mitigate harm could be inadequate. The academic literature is home to a large market of ideas for visual and auditory cues AVs could use to communicate to pedestrians. These include screens with text messages, images of eyes in a windshield, and different combinations of lights and sounds to warn individuals about an incoming action. In addition, researchers have evaluated several efforts with respect to their inclusiveness. Specifically, they have evaluated how individuals with different demographic characteristics and capabilities interpret the information displayed by AVs to safeguard human well-being.
2. Interaction with Authorities
Personnel representing certain public authorities (e.g., first responders, police, and firefighters) have privileges over roadways. In some scenarios, these individuals are charged with altering standard traffic rules to improve the flow of vehicles during an emerging situation. To fulfill their duties, it is evident that the capability to interact with AVs should be widely available and trained for. In particular, these individuals might require AVs to perform a specific set of actions for an unknown amount of time. Such circumstances can be faced in a variety of conditions such as construction work, torrential rain during a hurricane, a medical emergency, a riot, or a situation involving a firearm.
In addition to receiving orders from public authorities, AVs may interact with civilians who assume the role of informal authorities during accidents or unusual road conditions. For example, a pedestrian or another driver may want to modify the flow of traffic to protect an individual or a group from harm. In all cases, AVs must be able to follow instructions from formal and informal authorities. Equally, informal authorities ought to know how to perform their duties.
In terms of U.S. policy, the federal government acknowledges the desire for law enforcement to interact with AVs in its proposed model template for states. This document declares that “there will be a growing need for the training and education of law enforcement regarding their interaction with drivers/operators in both the testing and deployment of these technologies.” These interactions currently generate inconveniences for public officials. To address the above, entities testing and deploying AVs have developed law enforcement interaction plans. In states such as California, these plans are mandatory based on a vehicle’s permit, while in other states no such requirements exist. Depending on the AV developer, these documents provide important information that allows first responders to contact support personnel or disable, open, move, or tow an AV.
3. Interaction with Non-Road AVs
This Article has thus far described scenarios concentrated on road-faring AVs. An emerging market segment in the United States is one in which vehicles autonomously deliver goods using sidewalks. They generally exist in the form of mobile rectangular boxes that share pedestrian spaces while driving at slow speeds. Because of the setting, these AVs are in close contact with humans. However, unlike road AVs, there is clarity about the lack of a human physically in the driver seat of these sidewalk AVs.
Relevant in this scenario is the distribution of transit space that was mainly devoted to foot traffic. Therefore, the sidewalk is a context where pedestrians might be less accustomed to interacting with a technology that is crowding out a public space to complete last-mile deliveries. Here, communication is targeted at two entities. Particularly during crosswalk interactions, pedestrians and vehicles (both AV and non-AV) need to clearly understand a non-road AV’s intended action and direction.
Local governments throughout the United States have generated hard law to manage how sidewalk space is managed or shared with AVs, in some cases denoting delivery robots as “pedestrians.” More importantly, soft law initiatives in the form of ISO 4448 are available to standardize the behavior of this technology and are used as a reference for deployers and government stakeholders.
B. Infrastructure Support
AVs need sensors (cameras, radar, LiDAR, ultrasonic, etc.) to understand and interact with their surroundings. In addition to enabling autonomy, this technology generates data that allows the coordination of vehicles for the benefit of a user, a group, or society at large. Realizing the advantage from these capabilities hinges on interoperability. Without interoperability, each AV is limited to function as an independent entity incapable of distributing real-time information to cooperate with other systems. This Section discusses two scenarios for using the information generated by AVs to improve traffic flows through platooning and increasing the situational awareness of road conditions for public authorities. Both outputs have been widely touted by U.S. policymakers as examples of potential benefits from AV technology.
1. Platooning of AVs
Because they act individually, non-AVs are chaotic and inefficient in terms of group navigation and energy usage. In contrast, AVs can exchange information and coordinate their movements to optimally allocate space under current road conditions (this is called platooning in the literature). Platooning involves the synchronization of sensors and the communication of data. As AVs enter the U.S. market, there is an opportunity to harmonize the exchange of information as networking (e.g., 5G or 6G) or vehicle hardware (e.g., Lidar) technology improves. In this respect, there are soft law mechanisms attempting to set common protocols. The European Commission created the ENSEMBLE project to pool manufacturer expertise, test systems, and roll out platooning among heavy-duty vehicles. After four years of research and development cooperation, ENSEMBLE “created a common solution ready for standardization” that enables different vehicle brands to communicate and platoon. Internationally, standards and protocols have been developed to securely communicate vehicle-to-vehicle platooning data, such as one developed by Feifei Liu and her coauthors to relay information through cellular networks.
2. Data for the Public Benefit
Gathering information from the physical world for AV use requires significant infrastructure. For instance, several metropolitan areas in the United States employ dedicated sensors to monitor road conditions, while a minority, like Los Angeles, invest in dedicated hardware to optimize traffic. For governments of diverse sizes and budgets, the sensors and communication capabilities of AVs are an opportunity to outsource the systematic gathering and transmission of information on road conditions. The public benefit of these sensors need not stop at the gathering of data on road conditions. The combination of demographic, geographic, and meteorological information, among other sources, can help public officials concerned with improving their community (e.g., its health outcomes). Although this prospect sounds beneficial, procuring such data without carefully considering limits on its use could provoke direct and indirect harms to individual rights and civil liberties.
For AV data to become a net positive contribution to the public, there are a few issues to consider. Manufacturers need to establish interoperability between their systems to capture information and relay it to authorities. Associations representing local departments of transportation and AV manufacturers can employ soft law in the form of standards to describe the linkages needed to send and receive data. In addition, priorities on what type of information to classify, catalog, and distribute must be considered as a multi-stakeholder process that can be formalized through soft law. Agreements can be achieved on having AVs assist in generating reports on accidents, potholes, micro-weather conditions, natural disasters, and energy use. However, the monitoring and characterization of individual behavior could lead to major disagreements on how to balance privacy and the recovery of information that should be used exclusively for the public benefit.
C. Information Security
Without data protection, any effort to communicate between and with AVs is subject to a long list of attack vectors that can exploit supply chain vulnerabilities, remotely disable fleets, disrupt sensors, facilitate theft, and so forth. For users, inadequate data protection can lead to the loss of privacy and threats to physical security. Furthermore, a compromised information network is a clear indicator of a technology that cannot be trusted by consumers to perform any of its expected or desired tasks related to safety, infrastructure support, or profit motive.
To combat potential attacks, stakeholders have access to several soft law mechanisms. For one, the National Highway Traffic Safety Administration (NHTSA) dedicates significant bandwidth to educate the transportation industry about cybersecurity through workshops and publication of a best practices document. Industry alliances, like the Alliance of Automobile Manufacturers and the Association of Global Automakers, have created consumer privacy protection principles to align industrial efforts. Over twenty firms endorsed these principles, which were created in 2014 and altered in 2018 and 2022.
A different realm of soft law action relates to standard setting organizations. These entities exist in several forms, from professional associations, such as the Institute for Electrical and Electronics Engineers (IEEE), to state-centered organizations, such as the International Organization for Standardization (ISO). Committees of volunteers (in professional associations) or representatives (in state-centered organizations) develop standards for both technical and nontechnical fields. Within the AV information security sector, ISO/SAE 21434:2021 is a standard specifying “engineering requirements for cybersecurity risk management regarding concept, product development, production, operation, maintenance and decommissioning of electrical and electronic (E/E) systems in road vehicles, including their components and interfaces.”
D. Profit Motive
The safety, infrastructure support, and information security dimensions of this framework dealt with AV capabilities with a primary motivation that contributes to the common good (e.g., improve traffic and minimize accidents). This dimension emphasizes incentives that are not altruistic and instead are centered on the monetization of AV capabilities for the benefit of individuals or firms. Its inclusion in this framework stems from the desire to catalyze debate on the guardrails necessary to balance profit motivations with respect for civil liberties.
AVs function inherently as data magnets. Their sensors capture, process, classify, and transmit their surroundings to maneuver through the world. What if these sensors were given the ability to monitor data without a navigational purpose? Potential secondary targets could serve an almost infinite number of objectives. For example, market research firms could identify the demographics of pedestrians in specific geographies or count the number of vehicles in a business parking lot. Less desirable objectives include having a private investigator purchase access to this stream of information to surveille an individual or group.
In the United States, the First Amendment protects the ability of individuals and firms to collect information in public. Arguably, this includes data gathered by AVs through their sensors. Enacting hard laws to control this information intake and processing may result in legal controversies that, eventually, have no effect in deterring this activity. In this scenario, soft law may represent an effective alternative by setting norms that limit such surveillance. Importantly, prior efforts related to this type of monitoring have faced difficulties. Despite the legal hurdles to control the behavior of private actors, soft law can serve as a mechanism to limit unwanted outputs of a profit-motivated actor. For instance, the public can reveal preferences against behavior that generates some sort of automotive red-lining. By expressing these social norms and translating them into consumer choices, stakeholders can generate incentives against behaviors that pose risks and harms that would be difficult to regulate or take a significant amount of time to fight in court.
II. Governing AV Communication Through Soft Law
The framework proposed in this Article was developed to catalyze governance conversations on the implications of information exchanges by AV technologies. The framework’s objective is to tackle the pacing problem from increasingly advanced AV communication capabilities and their management by developers, users, and policymakers. To this end, each of the framework’s dimensions either recognizes existing uses of soft law or highlights opportunities to implement this governance lever in activities that merit sectoral stakeholder attention (see Table 1).
