When flying in an airplane, are you really able to tell when the captain is in control or when autopilot is engaged? Perhaps not.
So why are people slow to trust self-driving cars? It could be that we expect self-driving cars to be perfect, albeit an unrealistic expectation, says Kevin McFall, associate professor and interim chair of Kennesaw State University’s Department of Mechatronics Engineering.
“There are always going to be vehicle accidents, even 50 years into the future,” says McFall, who researches self-driving vehicles and machine learning. “However, a self-driving car will never drive drunk or distracted, never speed to impress friends and never run a red light because it is in a hurry. When you take all of this into consideration, I believe safety will be much improved overall with self-driving car technology.”
The technology still has a way to go. Just last year, ridesharing company Uber made headlines when one of its self-driving vehicles killed a pedestrian, and last year a Tesla was involved in a fatal accident with its Autopilot mode activated. Though it doesn’t offer full self-driving capability, Tesla’s Autopilot is a glorified lane assist program that doesn’t require driver input. It doesn’t operate the entire vehicle on its own. Regardless, just because an accident occurs doesn’t mean the ability of self-driving cars is invalid, and oftentimes the artificial intelligence controlling the vehicle is learning from its mistakes, McFall says.
In short, your car is studying your driving habits in order to perfect its own ability.
“A major component of machine learning is collecting enough data points to make the machine – in this case a car – more reliable,” he says. “The Tesla you own is sharing your driving data and, in return, the company is getting a treasure trove of information that allows them to make the system better.”
Autonomous vehicles have remained mostly in the testing phase in a handful of cities nationwide, but the technology is now being used on more simple, routine routes in contained areas. For example, Peachtree Corners, a suburb of Atlanta, is setting up an autonomous bus service along Technology Parkway. This is how self-driving vehicles should roll out, McFall says, because it allows companies to build public trust and familiarity before expanding to personal vehicles.
“We’re starting to see trucking companies and bus lines move in this direction because they are in a more contained environment,” he said. “Self-driving vehicles are much more reliable when they’re driving the same route every time. It becomes more complicated when the destination changes for each trip.”
So how soon will it be that commuters can work while stuck in traffic? Better yet, will there even be heavy traffic once autonomous vehicles become the norm? Unfortunately, it’s hard to predict how long it will take for the technology to become widely trusted because automotive companies keep their data close to the vest, McFall said. However, there are some undeniable benefits should self-driving cars continue to become mainstream.
“Even if congestion is unaffected, productivity savings of being able to work in transit are immense and will most certainly become reality soon,” he said. “It requires no communication or coordination with other vehicles. However, the truly transformative possibilities are linked to inter-vehicle communication. Imagine if all vehicles began moving at once when a street light turns green.”
So when it comes to self-driving cars, there will always be some risk associated with using the technology. But will the risk ever outweigh the potential benefits? It’s doubtful, McFall says.