Prepping Cities for Vehicle Autonomy

Deployment of new V2I technology and upgrades to existing roadway infrastructure can assist AVs and improve safety for all road users.

Ford is working with Argo AI to help equip cities including Austin, Texas, with technology for its impending deployment. (Argo AI)

While automated vehicles (AVs) use existing streets, highways, traffic lights and signage for guidance, they interact with roadway infrastructure much differently than human drivers. Because AVs are driven by machines instead of people, they have the advantage of communicating with other machines that are already in place and are starting to be deployed as part of smart infrastructure. Ranging from traffic cameras to sensors, collaborative infrastructure can help AVs navigate a tricky intersection or even detect and prepare for a potentially dangerous situation ahead, such as a pedestrian walking into traffic or a vehicle speeding through a red light.

Ford installed ‘infrastructure nodes’ outfitted with sensor arrays at intersections in Michigan and Florida that provide a birds-eye view of the intersection to augment AVs’ ground-level observation of an intersection. (Ford)

While most AVs don’t have to rely on machine-to-machine communication, also known as vehicle-to-infrastructure communication (V2I), the technology not only aids AVs but improves safety for all roadway users. “It would be wonderful to have infrastructure available to enable automated driving,” said Kay Stepper, senior VP of driver assistance and automated driving for Bosch. “But at this point it’s augmentation. Once we have infrastructure, we’ll embrace it with open arms,” Stepper continued. “We want stationary objects and moving objects to communicate. But we at this point we don’t have the penetration in the field.”

Using existing traffic cameras

But the situation is quickly changing. In many instances, the hardware is already in place, including thousands of traffic cameras. “Cameras have been used for traffic management for decades,” noted Frank Sgambati, a director of business development at Bosch, who specializes in smart city solutions. “For years, people have been waking up and checking traffic cameras to see how their morning commute will be,” he added.

“Now we’re using video as a sensor to classify vehicles, pedestrians and bicyclists at intersections,” Sgambati said. “Bosch cameras can now provide metadata and analytics in addition to images and we’re using that to determine – just like an AV’s sensors and software – whether a pedestrian or a bicyclist is in the roadway.”

For example, Bosch, the Ohio DOT and other partners have developed a Smart Mobility Corridor along US-33 near Columbus. The 35-mile (56-km) stretch of smart roadway runs through cities such as Marysville and Dublin that also are partnering with Bosch through Drive Ohio, the smart-mobility division of the Ohio DOT. “It’s a test bed to use video as a sensor to augment automated and connected vehicles and make automated driving safer and more seamless,” Sgambati said. “It’s really foundational work.”

In Marysville, three intersections have Bosch cameras as well as traffic-infrastructure supplier MH Corbin’s CONNECT:ITS system that takes information from cameras and sends it to vehicles using Dedicated Short Range Communication (DSRC) technology. “Vehicles that have DSRC capability can get information before they approach an intersection,” Sgambati explained. “That’s an early example of that technology.”

Bosch also is working on a similar project with Michigan DOT along a 7-mile (11.3-km) stretch of Detroit M1 highway, also known as the Woodward Corridor. “We’re installing camera-based smart infrastructure at about 57 intersections,” Sgambati noted.

In a similar vein, Ford- and VW-backed AV company Argo AI have partnered with Rapid Flow Technologies to use traffic cameras at intersections to collect data, including the number of vehicles, pedestrians and other road users approaching a traffic light. Rapid Flow Technologies’ Surtrac AI platform generates predictive models that allow Argo vehicles to not only react to other road users in an intersection but also predict their behavior.

AV company Motional, a joint venture between Hyundai and Aptiv, recently initiated a pilot with Derq, which supplies real-time images from traffic cameras and data from roadway sensors to government agencies and mobility companies. As part of the pilot, cameras positioned above busy intersections and connected to Derq’s roadside computers transmit data to Motional’s vehicles.

Derq’s data provides Motional AVs with a view of an intersection that their onboard cameras and sensors may not normally see and can help an AV detect pedestrians crossing a street behind parked cars or cyclists weaving through traffic at a red light, for example. Derq’s technology also can predict movements of other road users to alert Motional’s AVs of difficult traffic situation before a vehicle enters the intersection.

Smart traffic lights

AV companies also are working with cities to install new transportation infrastructure, mainly at intersections. For example, Ford is working with Argo to help equip cities with technology for its impending deployment in Miami, Florida and Austin, Texas.

To help prepare for Argo’s deployment in Miami-Dade County, Ford installed “infrastructure nodes” the automaker designed in-house and installed at two intersections in the Detroit suburb of Saline, Michigan. Equipped with cameras, radar and lidar sensors and perched above the intersection, the nodes provide a birds-eye view of the intersection to augment an Argo AV’s ground-level observation of an intersection.

A Ford-Argo animation capture showing the technical challenges by road intersections and how infrastructure can play an important role in ensuring traffic safety. (Ford)
Three intersections in Marysville, Ohio, are fitted with Bosch cameras and a “Connect ITS” system from traffic-infrastructure supplier MH Corbin. The information captured by camera is sent to vehicles using DSRC. (MH Corbin)

Ford has since installed the same type of smart infrastructure nodes in Miami Beach to communicate with Argo test vehicles at a particularly complex intersection. The intersection at Lincoln Road and Lenox Avenue in Miami’s South Beach consists of narrow two-way streets with buildings and palm trees close to crosswalks. It’s adjacent to a popular outdoor mall with a constant stream of cars, pedestrians and bicycles.

The intersection is difficult enough for human drivers to navigate and could create blind spots for the sensors on an AV. If an AV sensor’s line of sight is obstructed because of a building, tree or large truck, the smart node can provide information on what’s blocked from the AV’s view, helping it to react accordingly. The smart node also can provide a comprehensive view of the intersection before an AV even arrives, so that it can process the information and prepare for driving through.

Before Motional launched a public AV pilot with Lyft in Las Vegas in 2018, the company worked closely with Las Vegas-area transportation agencies to install DSRC units to allow Motional vehicles to “talk” to traffic lights and other roadway infrastructure. To date, Motional has equipped more than 120 intersections with DSRC technology that can be used by not just by Motional AVs, but any smart, connected vehicle.

Map and infrastructure updates

One vexing problem for AVs is temporary road construction and sudden changes to infrastructure. Argo AI works with the eight cities in which it operates, including Munich and Hamburg, to get updates on construction info. “We work with Miami to update us on road construction and build it into our 3D maps,” said Alan Hall, director of communications at Argo. “When cities are making changes to the road, we can block off that area on the map.”

Argo’s maps also are updated when the company’s vehicles come across changes in roadway infrastructure. For instance, if an Argo vehicle’s sensors detect a new stop sign, it archives this information and shares it with the entire Argo fleet. Argo also is in discussion with its city partners on the placement of new signs and infrastructure, Hall noted, so that the company can be proactive about adding these to its maps.

While V2I communication may be lagging AV technology, self-driving vehicles are starting to be deployed without this extra support. “AVs currently have the sensors and compute power needed to deploy,” said Bosch’s Stepper.

Argo has tested its vehicles using V2I technology. The company views it as additional sensing input that engineers can tap into, according to Hall, ,“We’re developing our self-driving technology to operate as the world is, not as we’d like to be,” he asserted.

While AVs don’t require V2I technology to operate, it can make the road safer for everyone, although it comes at a cost for cities. Fortunately, the U.S. DOT and state DOTs have been providing funding for smart-traffic infrastructure improvements, such as with Ohio Smart Corridor. The new, $1-trillion federal infrastructure bill also is expected to spur investment in the technology.

And by using existing traffic cameras and infrastructure, cities and other government agencies don’t have to invest in expensive technology, especially for a new source of transportation that’s not yet proven. “Infrastructure takes a long time to invest and change,” said Rob Spillar, director of the Austin Department of Transportation. “We’re used to building transportation infrastructure that lasts 60 years. “To have something last a few years, that’s one of the hardest things working with technology companies.”