The race is on. Tech giants like Google, Tesla and most car makers — from Toyota to Ford and Volvo — are betting billions to get intelligent self-driving cars on the road as soon as possible. Earlier this month, Apple became the 30th company to be granted permission to test autonomous vehicles on California by the state’s DMV. Even Amazon is reported to have created a team to understand how to best leverage the upcoming revolution. And Waymo (sister company of Google) has started testing its fleet in Phoenix.
But the road to autonomy is long, winding, and dangerous. Hours after being deployed in San Francisco in December, Uber’s autonomous cars were caught running a red light. Last year in Florida, a Tesla driver died in a crash when the car, on autopilot, confused a truck making a turn for an overhead road sign.
“I don’t see the ideal of complete automation coming anytime soon,” Christopher Hart, former president of the National Transportation Safety Board, said recently. “There’s no software designer in the world that’s ever going to be smart enough to anticipate all the potential circumstances… the dog that runs out into the street, the person who runs up the street, the bicyclist, the policeman or the construction worker,” he said.
So are we stuck with only humans as drivers for years to come? I don’t think so. But if we want to take our hands off the wheel anytime soon, we should stop focusing exclusively on the car. Instead, we should build smart lanes in highways dedicated exclusively to self-driving cars, far from any erratic human behavior.
Here’s how it would work. First off, we should zero in on a stretch of highway already equipped with intelligent sensors and road signs. We should then cordon off one lane with guard rail and make it accessible only to vehicles with the required technology (360° sensors, radar, vehicle-to-vehicle radio, automated steering, adaptive cruise control), through something like an E-ZPass toll booth. As vehicles enter and exit the lane at designated checkpoints, the system would assign each car a slot in the “platoon.”
The convoy would cruise safely at high speed, each vehicle “tailgating” the next. Through on-board radios, each car and highway sensor would be a part of the same moving network, routing around obstacles and becoming aware of disruptions far ahead of time.
Cars would save fuel (only the platoon leader would face headwinds), and drivers — or rather, passengers — could sit back and relax.
The technology is already there. Protocols allowing cars to wirelessly communicate with each other (V2V) and with infrastructure (V2I) will soon be mandatory for all new vehicles, and WaVe, a technology similar to Wi-Fi, will soon be embedded in traffic lights, stop signs, and work zones, allowing two-way communication between cars and infrastructure.
The idea of laying out vehicles in platoons is also not new: Swedish truckmaker Scania announced recently it will deploy autonomous truck platoons along a route in Singapore, while three Mercedes trucks platooned for 375 miles from Stuttgart to Rotterdam.
Tech companies often have little patience for working with regulators: They like to build, ship, test, and iterate, pushing legal boundaries where needed, like Uber is currently doing. But tech slogans like “move fast and break things” work a lot better with social networking apps than with the physical world, where the breakable things are humans.
Intelligence is proportional to the scope of the application, and city roads contain infinite variables. They are best navigated by highly adaptive (if easily distracted) brains, not machines that can only handle predicted scenarios.
Today’s intelligent cars will not be able to handle urban roads and unpredictable humans for years to come. But they could safely self-drive for hundreds of miles on smart highway lanes starting today, creating a crucial stepping stone for broader adoption in less controlled scenarios.
So how much would this cost? A useful comparison might be with toll lanes: Depending on complexity and length, new toll lanes can cost from a few hundred million to several billion dollars, often financed through joint public-private partnerships.
That might seem like a lot — until we look at the economic impact automated vehicles will have on the U.S. economy, which University of Texas engineers have estimated at $1.2 trillion. When a company like Intel spends $15.3 billion in one acquisition to achieve leadership in the connected car industry, it’s hard to see why U.S. federal and local governments could not spend a fraction of that to help build the future of transportation.
Should President Trump fulfill his promise to allocate $1 trillion to improving U.S. infrastructure, there might even be a budget for it.
Umberto Malesci is the cofounder and CEO of Fluidmesh, a New York and Milan-based wireless communication systems vendor that helps trains, robots, video surveillance systems, and other “things” connect securely.