Highly automated driving systems (ADS), often referred to in the media as “self-driving cars,” present challenges and opportunities for public safety. In 2017 alone, motor vehicle crashes caused 37,133 fatalities in the United States.¹ Additionally, 94 percent of crashes are attributable to human error.² Given these factors, ADS technology has the potential to significantly improve safety on U.S. roadways, as well as to increase traffic efficiency, economic opportunities, and access to mobility for some individuals who currently face transportation challenges.

However, in order to fully realize these benefits, ADS-equipped vehicles will first need to be able to reliably navigate a vast number of complex and dynamic situations. Operating scenarios requiring complicated and less predictable interactions with public services or those that involve rapidly changing traffic conditions remain challenging for ADS to navigate safely and predictably.

As an example, consider a scenario in which an ADS-equipped vehicle with no front seat occupants approaches a freeway crash. The lanes are blocked, and a police officer is directing traffic. Today, this interaction would likely be accomplished through eye contact and hand gestures between the human driver and the law enforcement officer. However, in the case of an automated vehicle, questions of how to substitute for this human interaction need to be considered to ensure the safety of vehicle occupants, first responders, and other road users. Would the vehicle sense and respond as the officer expects to hand gestures or a whistle? Would the ADS correctly interpret the context of a specific interaction, particularly considering the varied strategies employed by first responders while directing traffic? How will the officer know if the vehicle understands the instructions, and how can the vehicle communicate other pertinent information to the officer (e.g., if the vehicle is not able to drive the new path requested by the officer)? These are just a few of many challenging scenarios for which developers, infrastructure owner-operators (IOOs), and the public safety community must collaboratively resolve to enable public trust and bring to fruition the transformative safety benefits of this technology.

To this end, a research study, funded by a $7.5 million grant from the U.S. Department of Transportation (USDOT) and another $3 million in contributions from the project partners, seeks to develop solutions and demonstrate how ADS-equipped vehicles can interact safely in these scenarios. The project will

• • • define key dynamic scenarios;
    • develop technological concepts for safe interactions; and
    • demonstrate the solutions in a Northern Virginia corridor optimized for automation while collecting data for ADS development, safety analyses, and rulemaking consideration.

Led by the Virginia Tech Transportation Institute (VTTI), the demonstration team includes vehicle manufacturers representatives from the Crash Avoidance Metrics Partners LLC (CAMP); infrastructure owner operators (IOOs) Transurban, operators of the express lanes for interstates 495, 95, and 395; the Virginia Department of Transportation (VDOT); and numerous public safety stakeholder representatives from across the United States, collaboratively addressing these challenges and creating broadly deployable solutions.³

Implications for Public Safety

Public service providers will be critical in integrating ADS into the U.S. transportation system. Once the systems are fully deployed, public service representatives such as safety officers, emergency medical personnel, and construction workers could, with the appropriate knowledge and training, safely operate daily alongside ADS-equipped vehicles. Providers will need to know (1) how to interact with the ADS and (2) how they should expect ADS-equipped vehicles to respond in critical situations along the roadway.

Collaboration is a foundation of this project, which brings together some of the largest national voices within the public safety, original equipment manufacturer (OEM), and IOO communities. The demonstration approach and team composition are a direct result of extensive discussions with these groups, which revealed some of the most essential safety challenges that need to be addressed. These include scenarios in which high-level ADS must respond appropriately and safely to a range of dynamic conditions, including lane closures and merges associated with work zones on freeways and arterials, and interact directly (i.e., through physical contact) and indirectly (i.e., communicate without physical contact) with public services personnel.

These scenarios are characterized by interactions with infrastructure or other actors (cars, workers, etc.) that tend toward the unexpected or atypical, including interactions with public services. Many of the features necessary for reliable perception and response in these scenarios may be obstructed, incorrect, or difficult for ADS to identify and interpret at this stage of development. This presents a safety concern for users of ADS-equipped vehicles, other drivers sharing the roadway, and public service providers. In addition, responders and workers operating within these scenarios are already often at risk due to the combination of their proximity to traffic, lack of physical protection, and the complex nature of the environments and situations they work in. It is critical that ADS do not exacerbate the risks already faced by these workers; rather, the research team is striving to improve safety for the individuals performing these valuable services to the community.

Across the United States, these scenarios represent thousands of interactions per day and, thus, must be addressed in order to achieve a safe and efficient transportation system. With extensive input from OEMs, IOOs, and public safety stakeholders, the research team will identify targeted dynamic scenarios, define the interactions between the actors and ADS, and develop technological solutions.

Demonstrations

Following the development of technological solutions, the VTTI team will build reference vehicles with advanced ADS capabilities to demonstrate how coordinated interactions between infrastructure, public safety, and automated vehicles can enable safe dynamic traffic scenarios on the I-95 Express Lanes in Northern Virginia. Managed lanes, such as Transurban’s Express Lanes, provide an ideal place for ADS testing and demonstration since the lanes are closely monitored, include supporting infrastructure technology, and are separated from general traffic. The demonstrations will include a combination of controlled and naturally occurring scenarios coordinated in collaboration with IOOs, public safety organizations, and OEMs. To maximize impact and educational potential, the demonstration events will be captured on video and disseminated through industry and media groups.

The vehicles will be equipped with a prototype reference ADS package that also includes the ability to exchange data through connected vehicle systems. This capability will allow the ADS to cooperate with the roadway when needed for the ADS-equipped vehicle to maneuver in the selected scenarios. The scenarios will be extensively validated and refined, with consideration of various interaction strategies and the practicality of technological solutions, while maximizing performance and knowledge acquisition.

Data Collection

Comprehensive data will be collected during testing and demonstrations:

• • • Raw data from various perception sensors (e.g., inertial measurement units, LiDAR, radar, and GPS)
    • Infrastructure data, including messages exchanged between ADS and infrastructure and messages exchanged directly between scenario actors in some scenarios
    • Survey and focus group responses reflecting the perspectives of targeted stakeholder demonstration participants (e.g., transportation-challenged populations and first responders).

Collected data will be processed into discrete data epochs for a set of collaboratively selected complex scenarios demonstrated in real-world settings. These data will be shared openly to allow ADS developers to investigate the scenarios through the “digital eyes” of their ADS. The data will also permit the research team to investigate ADS readiness at the interaction layer. In addition, key data elements will be abstracted into a powerful 3D simulation environment, enabling countless additional scenario investigations through modeling using industry-standard toolsets. This simulation environment, focused on complex dynamic scenarios, will be extremely valuable for safety analyses and will serve as a training tool for an outreach program, providing the industry and public services with a common understanding of how to appropriately facilitate safe interactions within these scenarios.

Figure 1. Example of a simulated driving scenario (left) created by GCAPS based on an actual driving scenario captured in forward video within VTTI’s naturalistic driving data (right).

Finally, the team will analyze the data with respect to the effectiveness of ADS implementations across the scenarios tested. This will include evaluations of the following elements:

• • • The ability of ADS, with appropriate technological solutions, to successfully navigate complex dynamic scenarios as intended
    • Stakeholder feedback to validate strategies and identify opportunities for improvement across each scenario
    • Readiness assessments by representative groups from across the United States that will interact with ADS in the future (e.g., police, fire, OEMs, maintenance crews, and transportation-challenged populations)

The research objectives outlined will be carried out over the course of a three-year period of performance.

Previous Research

As a leader of advanced vehicle technology evaluations, the project team is continuing its efforts to solicit valuable insight and feedback from public safety stakeholders with respect to the deployment of automated driving systems.

Prior work conducted by VTTI, in a collaborative effort funded by OEMs represented by a CAMP Consortium called Automation and Public Safety Common Solutions, highlighted opportunities for ADS-equipped vehicles to improve safety for Emergency Services personnel.⁴ As a key part of that effort, public safety stakeholders were given an opportunity to share their understanding, concerns, and requests surrounding ADS-equipped vehicle capabilities. VTTI analyzed the operational tasks of public safety personnel in scenarios consisting of a high level of interaction with vehicles of the public, including responding to an emergency incident, securing a crash scene, and conducting a traffic stop. Through analyzing these key operations specific to law enforcement, the research determined how interactions at the task level would be affected by ADS-equipped vehicles with and without occupants. Analyses included a focus on interactions that incorporated direct or physical contact with a vehicle, solicitation of information from a driver (e.g., request for license or registration), and signaling of information (e.g., lights, sirens, hand gestures). Additionally, CAMP has completed several other pre-competitive research projects on cooperative automation and traffic optimization, including the concept of improved freeway traffic flow.

The results of this work provided a rare opportunity for public safety services to have a voice in the early process of ADS development. The task analysis revealed potentially significant gaps with interactivity between public safety and ADS-equipped vehicles, thus providing the opportunity for those gaps to be addressed with technological solutions. Now, in moving forward with the development of reference vehicles equipped to navigate the to-be-selected dynamic scenarios, the research team is aware of key situational concerns facing public safety personnel and can leverage these insights to facilitate the advancement of emerging technologies as solutions to enhance safety.

Conclusion

The research team is targeting some of the most complicated and difficult challenges faced within the ADS industry today. Thus, the demonstrated solutions, along with the datasets and outreach materials produced by this project, will be extremely valuable to a variety of public service providers. Throughout the project, the team will engage providers across jurisdictions and domain types (rural, urban,  etc.), along with additional public service stakeholders. Their representation will ensure that public service providers come away with a better understanding of how ADS will respond to them and how their multijurisdictional community can proactively prepare for ADS deployment. In turn, ADS developers will be more likely to apply the developed solutions because they will have been vetted within the public services community. Ultimately, this project strives to produce the basis for standard protocols covering complex interactions with ADS, focusing on improved communications between public officials and automated driving systems.

Mike Mollenhauer is the interim director of VTTI’s Technology Development and Deployment.

Zachary Doerzaph is the director of VTTI’s Center for Advanced Automotive Research and a faculty member within Virginia Tech’s Department of Biomedical Engineering and Mechanics.

Travis Terry is a research associate for VTTI’s Center for Infrastructure-Based Safety Systems.

Notes:

¹ “2017 Fatal Motor Vehicle Crashes: Overview,” Traffic Safety Facts (Washington, DC: National Highway Traffic Safety Administration, 2018).

² “Critical Reasons for Crashes Investigated in the National Motor Vehicle Crash Causation Survey,” Traffic Safety Facts (Washington, DC: National Highway Traffic Safety Administration, 2015).

³ Crash Avoidance Metrics Partners (CAMP) LLC;  Transurban; and Virginia Department of Transportation (VADOT).

^{4 “}Automation and Public Safety Common Solutions (APSCS),” CAMP, 2018.

Please cite as

Mike Mollenhauer,Zachary Doerzaph, and Travis Terry, “Automated Vehicle Implications for Public Safety: Exploring Solutions for Interactions between Automated Driving Systems and Public Services,” Police Chief Online, September 2, 2020.