In Pennsylvania's big city to the west, autonomous vehicles with their black, high-tech conning towers are becoming a familiar sight.
"The first time I saw one I thought it looked like something out of Ghostbusters," said Francis Shirey, a 50-year-old Uber driver in Pittsburgh who, some day, could be put out of work by one of those funny-looking cars.
On Wednesday the ride-hailing app giant Uber, which has been testing its self-driving cars since the spring, began offering passengers in Pittsburgh free rides in a few self-driving vehicles, and it plans to have about 100 robo-cars motoring around the Iron City by the end of the year.
As someone who has sampled the technology firsthand, I can say it's both exhilarating and a little frightening. A week before Uber started offering rides in its cars, I took a trip in a similar autonomous vehicle built by Carnegie Mellon University.
Looking out the windshield of a self-driving Cadillac, I was peeking into a future that may be 10 years away, at most, where cars don't just drive themselves but talk to each other and their surroundings. They will cooperate to lessen traffic congestion, rather than drive aggressively and selfishly to make the situation worse (sorry, humans). Death by traffic accident may become a rarity.
Advocates say smart transportation technology will also bring more mobility for the elderly and blind and a better way to maintain our transportation infrastructure.
Someday the idea of driving your own car may seem as reckless as a transatlantic crossing in the age of the sail.
"This technology will revolutionize transportation as we know it," said Raj Rajkumar, a Carnegie Mellon University professor and co-director of the Connected and Autonomous Driving lab, of autonomous vehicle research.
But, he is quick to add, "we are not there yet."
Uber's self driving Ford Fusions are equipped with 20 cameras, seven lasers, a 360 degree radar and, always, a driver - just in case.
That's required by Pennsylvania law, but it's also a necessity.
The experimental state of autonomy quickly becomes apparent during an occasionally nerve-wracking drive in Carnegie Mellon's autonomous vehicle, a 2011 Cadillac SRX - which uses similar technology to Uber's vehicles.
It's equipped with six lasers, six radars and three cameras that scan the environment around it. An electronic hum emanates from the trunk, where four computers process data from GPS and preprogrammed road maps. On the dashboard is a screen showing the car's progress, with small pinpoints representing the outlines of obstacles the vehicle's lasers are detecting.
As the Cadillac pulls out of Bob O'Connor Golf Course near Carnegie Mellon's campus onto the open road, Rajkumar, in the driver's seat, talks and gestures as the wheel spins, as if directed by ghostly hands. The car recognizes a stop sign ahead and brakes, but when a minivan pulls out from a parking spot Rajkumar quickly turns the knob disengaging the robotic driver and grabs the wheel. A jarring reminder that the technology isn't 100 percent reliable even on mildly busy side roads.
There's a standard reaction when people first interact with an autonomous vehicle, Rajkumar said. For the first 10 minutes or so they're nervous. But then they get comfortable, and that's dangerous.
"After 10 minutes, the problem is they stop paying attention," Rajkumar said.
These vehicles rely on GPS maps and, in some cases, preprogrammed routes, but can run into problems if their cameras can't see the lines on the road, as happens if there's snow. Rain can confuse the system, and bridges are a problem. While crossing a bridge the car's lasers have nothing to bounce off, and it essentially registers as a void. If a driver isn't paying attention, as is thought to be the case in a fatal crash of a self-driving Tesla in Florida this summer, disaster can result.
Uber says having a driver in the vehicle and another staffer sitting in the front passenger seat to monitor the trips, offers enough security.
"We would not be using these cars if we did not believe they were safe," said Craig Ewer, an Uber spokesman.
Rajkumar, though, is wary of putting passengers into such cutting edge vehicles.
"Having employees of Uber exposed to risk is one thing," he said, "having the public exposed to risk is an extension of that."
Pittsburgh's mayor, Bill Peduto, is even more direct.
"It's not fail-proof" he said. "Nothing's fail-proof. There will be accidents."
Then he noted 1.3 million people worldwide die each year in car accidents. Autonomous vehicles may save many of those lives, and Peduto wants Pittsburgh to be the testing grounds for the transformative technology.
"The research that's been ongoing for decades is now at the point where the public has an opportunity to participate," he said.
Uber offers a more controlled introduction of the technology, said Daniel Lee, a University of Pennsylvania professor of electrical systems engineering who works with Carnegie Mellon researchers on transportation technology.
"They're trying to sell a service," he said. "You don't have to make the vehicle available for every single ride."
Uber's experiments are potentially just the beginning of what will become the new normal in transportation. Rajkumar and other CMU experts note partial autonomy is already here. Within two to three years, self-driving vehicles in controlled environments such as shipyards and airports could become the norm. Full automation for personal vehicles, Rajkumar predicted, is about a decade away. Lee, who said development will accelerate as private companies like Uber invest more in the technology, thinks 10 years may be a conservative estimate.
Pennsylvania is getting behind the technology, not just because of its potential to mitigate motor vehicle deaths, with 1,200 reported in the state last year, but because it could make the state a tech hub.
The state, and Pittsburgh in particular, offer excellent testing ground for the technology. Carnegie Mellon's expertise is a major attraction. Its robotics lab is so cutting-edge, Uber hired away 40 of its staff when it opened up its own Advanced Technologies Center in Pittsburgh last January.
Pittsburgh's geography and climate is also ideal for putting autonomous vehicles through their paces. Steep hills, more than 400 bridges, and chaotic urban roadways.
"We have four seasons, we have mountains, we have tunnels, we have bridges," said Kurt Myers, deputy secretary for driver and vehicle services at PennDot.
Uber's spokesman, Ewer, referred to Pittsburgh as "the double black diamond of driving."
Peduto noted the tech companies coming to the city, including Bosch and Google, have led to revitalized neighborhoods and thriving business districts, and the city's willingness to be a laboratory has allowed infrastructure upgrades with the bill footed by grants and corporate foundations.
In Pittsburgh, the cars won't just drive themselves, they'll talk to each other. Vehicle-to-vehicle communication offers the potential for WiFi "conversations" that can allow one car to alert others about road work, accidents and other obstructions, CMU researchers said.
The technology would also allow a wireless tether among a fleet of trucks and a lead vehicle, allowing them to "platoon," or operate in close proximity, and, like a cyclist saving energy by drafting behind another biker, drive more efficiently. Cars could also detect cell phones on pedestrians and react before a person even came into view.
The cars may also "talk" to street lights. Already in operation in Pittsburgh neighborhoods are 50 lights equipped with cameras and sensors that can detect where traffic is and adjust signals to pass vehicles through more efficiently. The existing smart signals have led to a 40 percent reduction in idling, which also leads to better air quality, said Chris Hendrickson, a CMU professor of civil and environmental engineering.
Further down the road are sensors on snow plows that can analyze the consistency of snow and advise drivers accordingly, and smart headlights that, instead of one bulb, are made up of thousands of LED lights that can adjust to highlight an animal on the road or reduce reflections off rain drops.
"You want the whole system to work together," Hendrickson said.
As smart transportation technology matures, engineering and software hurdles aren't likely to be the biggest obstacles to adoption. The stuff will initially be expensive. Right now, upgrading an intersection with smart traffic signals costs about $25,000, for example.
Even more challenging will be getting the public used to the idea of cars that don't need a human's touch. Peduto expected full adoption of autonomous vehicles to take decades because of people's reluctance.
Shirey, the Uber driver, was wary of machines taking the wheel.
"While we don't always trust people to do the right thing," he said, "we at least expect them to react like humans."