Ford’s Argo team brought a fleet of autonomous vehicles to the famously congested streets of Miami, in order to test AVs in some real-world situations.
MIAMI—There’s a lot to distract a driver in downtown Miami. On Biscayne Boulevard, a yellow Corvette is battling a red Mustang in a Fast and the Furious-style street race. Outside the Hilton DoubleTree, there’s a pack of bikini-clad young people at the curb, looking for their Lyft to the beach. Within the downtown inner loop, orange and white barricades block off streets and divert traffic with detour signs as construction workers build condos and repair roads in anticipation of hosting visitors for Super Bowl LIV in 2020.
But Ford’s autonomous car is taking it easy and staying on course. As we roll from the MiamiCentral transportation hub to Ford’s autonomous vehicle terminal tucked away at the southeast end of the mural-splashed Wynwood neighborhood, the self-driving Fusion sedan I’m riding in gamely navigates the maze of one-way streets, steers around construction sites, and waits patiently behind double-parked cars picking up passengers. When a shirtless pedestrian looks like he might step off the curb far from the nearest crosswalk, the car spots him and, acting on an abundance of artificial caution, brakes to a natural-feeling stop before the man even ventures into the street.
In the back seat, I watch the monitor mounted to the seatback that show a videogame-like scene of the road ahead and around the vehicle—a vision of what the array of LIDAR sensors atop the roof of the car are picking up. It’s easy to get lost in the digital renderings of cars and obstacles until you pull your head up and see the cyclist sidling up the Fusion’s blind spot on the screen materialize in real life. As the rider rolls past, the human backup driver in front puts her hands on the wheel, just in case—a reminder that, at this point in the self-driving-car saga, no one wants a safety hiccup.
The week before Thanksgiving, Ford invited a gaggle of reporters for a showcase of its AV program, dubbed “Miami Experience.” With makeshift AV-only curb lanes and mock-up of a ride-hailing iPhone, the event was designed to give the media a glimpse of what urban life in an AV-centric future city might feel like. For an afternoon, we hail Ford Fusions from pickup lanes to hop around town for an afternoon aboard vehicles that are mostly, if not exclusively, driving themselves. (In the front seats of each car, there’s a safety driver and an engineer on duty). Miami is set to be one of several proving grounds for Ford’s self-driving fleet, which will also be tested in Detroit, Pittsburgh, and Washington, D.C. The company says it plans to deploy 100,000 AVs across participating cities by 2021.
The rush to full autonomy has hit more than a few setbacks recently. After a self-driver Uber struck and killed a pedestrian in Tempe, Arizona, back in March, the ride-hailing company shuttered its Arizona testing hub and pulled its cars off the streets of Pittsburgh for several months. Tesla’s self-driving Autopilot feature recently logged 1 billion miles (10 percent of the fleet’s total mileage), but it’s also been blamed for several crashes, including one fatal crash in March, and Elon Musk’s 2015 prediction that a self-driving Tesla would go hands-free across America by 2018 is looking, like many such full-autonomy-related promises, to be too optimistic.
Meanwhile, Google’s Waymo, which had promised to launch a fully driverless fleet in Phoenix, has quietly put safety drivers back in its vehicles. General Motors says it will produce an autonomous ride-sharing fleet, Cruise, in 2019.
Amidst this fitful progress, AV skeptics and critics have continued to voice their concerns that self-driving will spawn any number of ill effects on cities, from traffic congestion and privacy issues to environmental and equity concerns. As of this week, a last-minute push from Congress could loosen up many federal regulations on developing and marketing self-driving vehicles, while also requiring crash information from companies and setting new standards for the developing technology that’s already on the roads.
Into this anxious and murky atmosphere comes Ford and its software subsidiary, Argo AI, which seems to be setting itself as the slow-but-steady competitor in the great AV race. The company brought its AV program to Miami in order to show off what a more harmonious relationship between a disruptive technology and a host city might look like.
Key to that idea is the company’s pledge to coordinate its rollout of AV-based services with cities, in order to have them better fit in. Its autonomous vehicle future will sync up and complement the city’s existing public transit, such as the already-autonomously operated downtown elevated Metromover that brought us to the launch event at MiamiCentral. The company is touting its City Solutions team, which will tackle community engagement with citizens and lawmakers, as one way to better steer future cars towards the goals of a city with sensitivity to the neighborhoods they enter.
And, as Ford CEO Jim Hackett tells me, the company is intent on proving to cities that its AVs won’t wind up being traffic-snarling urban menaces. “The confluence of the robotic vehicle and the cloud structure that allows these vehicles to communicate [with each other] will make it so there’s a hope in the future that when I get in a vehicle I’m not paralyzed by the congestion.”
Hackett compares the current unsettled state of the AV-verse to the world that greeted the very first iPhone users: The device itself may have been an amazing product, but it still needed improved cellular networks to actually function well. “People thought the iPhone didn’t work at first—you had to go in and out to take cellphone calls. But then the infrastructure caught up, and now you can go anywhere you want.”
Hackett thinks the same will ultimately be true with AVs and traffic congestion. “That same evolution will happen with the vehicle and the design of the system. That’s the promise here—that we reduce the congestion paradox as cities are growing.”
To accomplish that, AVs will have to not only get from points A to B without incident, they’ll need to figure out how to insert themselves into the automotive cultures of different places. “The culture of driving is a whole variable to the way humans learn,” Hackett says, comparing the difference between riding in a New York cab and driving in a small Midwestern town of 50 years ago. “The pace and the pulse of the city—that cab driver had a sense for that and they could move us through it.”
For a robocar looking to understand the vagaries of the human experience, Miami offers a lot to learn. The city has both the nation’s fifth-worse traffic congestion and a high ranking on lists for the most per capita pedestrian and bicycle fatalities. “The streets of Miami are exciting,” says Pete Rander, president of Argo AI, which develops the software that does the driving in Ford’s cars. “There’s so much to see, but they are challenging to navigate.”
As Ford’s AI gets to know Miami, its engineers are also testing other kinds of human interaction. For example, the car greets its passengers: “Hola! Jump in and relax. Leave the driving to me.” Adding that audio cue is something Ford gleaned by delivering pizza in the city with a Domino’s car outfitted to look and behave as if it were a self-driving car—one of its many Wizard of Oz-like self-driving simulation schemes designed to study how people will mingle with robot cars before they’re fully deployed.
Just how that deployment will unfold still remains uncertain, but Ford is spreading its bets around multiple business models and forging partnerships with all manner of mobility firms, from the microtransit service Chariot to the e-scooter startup Spin. The cheery electronic greeting suggests that the company expects AVs to pull fleet duty as ride-hailing vehicles, possibly through its partnership with Lyft. The carmaker also recently announced a deal with Walmart, building on its existing relationship with the courier app Postmates, that describes what sounds like a citywide conveyor belt connecting businesses with delivery services.
“Ford, unlike peers that have focused on just ride-sharing or fleet deployment, has the ability to try multiple business models,” says Adie Tomer, a fellow at the Brookings Institution’s Metropolitan Policy Program who has studied how AVs could reshape cities. “The transportation industry is huge. That’s why these companies are racing to test the market. But they can’t predict exactly what the business model might look like.”
One, somewhat dystopic AV business model would open up an alternative source of revenue by offering coupon-style coordinated stops on ride-hailing trips, turning them into roving advertising opportunities. Another idea offers busy parents the prospect of having a robot car pick your kid up from school. And look out pedal taverns: In a party city like Miami, AV services might coordinate with events venues to bring drinks and music to shuttle riders.
One of Tomer’s recent reports from Brookings estimates that 9.5 million people—or 1 out of every 20 workers—could see their work affected by autonomous vehicles. But this transformation won’t happen overnight: “It takes 20 plus years for the vehicle fleet to turn over, which speaks to the conditions and quality of the capital in invested in operating just metal vehicles,” says Tomer. “At the bare minimum, once autonomy hits, then you start a clock for 20 years for adoption that’s not even universal.”
Government regulation, at both the federal and local level, could affect the pace of AV adoption further: “We’ve seen the timelines move so much,” Tomer says. “It’s not because there’s reason to be skeptical of their veracity, but there so many other uncontrollable but important factors. Even if the tech is ready, will the cities be ready?”
If the company’s Miami event offered an occasionally convincing glimpse of what a city might feel like once self-driving cars ruled the streets, it also served as a reminder that, at the moment, neither the city nor the cars are truly ready.
At one point, the vehicle I’m in gets flummoxed waiting cautiously for an opening to make a difficult left-hand turn as cars blast past. Finally, the safety driver decides to take the wheel to execute the turn. On another ride, as we clear the second crosswalk of an intersection over Biscayne Boulevard, a confluence of traffic factors arrive all at once: A moped rider cuts a right hand turn on a red light, a bicyclist is riding past in the crosswalk, and another car is heading the opposite way into the intersection. While our car has the right-of-way, instead of proceeding it just stops for a few moments in the middle of the intersection to figure all these things out—something a real driver would be unlikely to do in a similar situation.
The delay is brief; the other cars behind us don’t even honk. But the pause still feels distinctly not human—one moment on an otherwise smooth and natural-feeling ride when the ones-and-zeroes behind the wheel made themselves known.
The engineers will study these incidents later, to better divine how to avoid them in the future. I ask our safety driver team about the stop, and whether other motorists ever get frustrated with an excessively careful self-driving car dithering in traffic. Yes, the co-pilot agrees, humans do sometimes get impatient with the pace of this technology’s progress.
“We’ve been honked at a lot,” he says.