Autonomous and connected systems have captured the imagination and are driving advances in many fields, from mobility to agriculture, robotics, space, and other areas. As the physical world, computing, and networking converge increasingly, it is important to distinguish between autonomous and connected as they relate to vehicles.
Although there is a lot of interest in the efficiencies and opportunities that can be gained through autonomous, or “driverless,” vehicles, there are huge near-term opportunities involving big data coming from connected vehicles driven by humans. This data can provide powerful insights to help government agencies improve the safety and efficiency of their surface transportation systems, and Purdue faculty are making big strides in using this connected vehicle data.
For example, we historically have used crash reports to identify roadway locations for potential safety enhancements. Connected vehicle data now provide a mechanism for agencies to obtain locations where hard braking occurs. In Indiana, we ingest information on about 12 million anonymized hard-braking events per month. Instead of waiting years for crash statistics to accumulate, we now can identify emerging hazards with just a few days or weeks of data. Furthermore, instead of relying on varying state crash-reporting processes, these techniques leverage connected vehicle data collected in an identical manner across the entire country, so the analysis techniques can be scaled nationally.
Focusing on things like this — which offer immediate benefit to agencies and the motoring public — builds important relationships between vehicle OEMs and public bodies. This data sharing, in turn, provides an engagement framework for some of the more challenging technical and public policy aspects of operating fully autonomous vehicles on our public roads.
Central to the advance of autonomous and connected systems is the idea of “edge computing,” a distributed computing strategy in which computational power and networking connectivity are placed at or near the spot — the “edge” — where they are needed, rather than data being whisked to a remote server for processing and back to the edge to execute an action. Edge computing is crucial when time lags between sensor awareness and action taken need to be measured in milliseconds — think collision avoidance in a dynamic, real-time autonomous driving scenario.
I think about vehicles themselves as examples of edge computing, in that they have extensive instrumentation to efficiently operate engines, windshield wipers, lane-keep assist, traction control, anti-lock brakes, and so forth. Much data on the operation of these features can be extremely valuable to public agencies. Also, looking at vehicles as edge computing naturally brings in critical partners in cybersecurity, sensors, Internet of Things (IoT), and communication networks.
Purdue colleagues are developing edge computing best practices for securely and anonymously sharing both real-time tactical information and off-peak data with public agencies. This data can be used for making real-time decisions, as well as prioritizing long-term capital infrastructure investments, for instance, in bridges, added lanes, and electric charging systems.
At Purdue, we’re looking to strengthen these partnerships, and the field in general, through the Purdue Engineering Initiative in Autonomous and Connected Systems (ACS), which has created a community of scholars to advance the science and engineering of autonomy, robotics and the IoT. ACS interdisciplinary teams are developing scalable architectures for deploying these technologies in a manner that improves safety and efficiency for operating our modern transportation systems.
Additionally, ACS members integrate their research into undergraduate and graduate educational programs, enabling students to gain classroom, real-world and international experiences, as well as access expert mentors. ACS faculty collaborate with external academic partners, corporations, and international organizations to promote and advance the state of the art in autonomy, robotics, and IoT engineering and science.
Although a lot of media focus is on autonomous vehicles on our highways, those are very challenging environments in which to provide robust operation, especially with variables like lighting and weather, and unpredictable interactions with human drivers or pedestrians. Our Purdue colleagues’ research in the autonomous agriculture equipment and unmanned aerial vehicle (UAV) spaces is providing critical stepping-stones for testing autonomous technology in more controlled environments, such as agricultural fields or structured air space. For example, in the last year, Purdue faculty have set up a center with the Indiana Criminal Justice Institute that has trained more than 60 agencies to program autonomous missions for drones to map motor vehicle crash scenes. To date, Purdue has processed more than 70 crash scenes.
I think we will see regular autonomous vehicle operation on rural interstate routes in a year or two. However, autonomous operation on dense urban interstates and in urban areas with significant pedestrian activity are several years away. Universities are very agile at assembling diverse funding portfolios of government agencies and private companies. For example, in Indiana, we process approximately 11 billion connected vehicle records per month that the Indiana Department of Transportation (INDOT) uses to continuously improve the safety of construction work zones. We also are leading a pooled fund study with 10 states (Texas, Ohio, California, Utah, Minnesota, Wisconsin, Pennsylvania, Connecticut, North Carolina and Georgia), using this data to develop traffic-signal performance measures that help agencies identify opportunities to reduce motorist delays and fuel consumption.
Historically, automotive companies have responded to government regulations — think seat belts, rear taillights, air bags, emission controls, and more. However, as we move into the autonomous space, it is clear that new collaborative relationships must be established between government agencies that build road infrastructure and automotive OEMs that build cars. Developing these relationships is critical so we can evolve our human-vision-centric view of road signs, pavement markings, and lighting to consider best practices that work with emerging vehicle sensors and autonomous technology deployed in vehicles.
Indiana has been a leader in this collaboration for several years. In March 2017, at the Purdue Road School Transportation Conference, INDOT Commissioner Joe McGuinness stated in his opening remarks: “Autonomous, connected vehicles are a thing of the future, and the future is now. We have to start planning and making sure that we are prepared for what the automobile manufacturers are going to be putting on our roads.”
Similarly, in his 2017 State of the State message, Governor Eric Holcomb said: “For Indiana, the Crossroads of America is more than a motto. It’s a mission” — signifying Indiana’s sustained commitment to work with all stakeholders to accelerate connected and autonomous vehicle deployment.
The Purdue Engineering Initiative in Autonomous and Connected Systems has made great strides over the last four years in addressing the challenges identified by Indiana leadership, and has established Indiana as a national leader in this space.
Darcy Bullock, PhD
Lyles Family Professor of Civil Engineering
Director of the Joint Transportation Research Program
Courtesy appointments in the School of Mechanical Engineering and the School of Electrical and Computer Engineering
College of Engineering, Purdue University
Video: Key concepts of “Deriving Operational Traffic Signal Performance Measures from Vehicle Trajectory Data,” named best paper by Transportation Research Board’s Traffic Signal System Committee in January 2021