If you ever find yourself at the San Jose airport hailing a cab to take you into San Francisco, prepare for your own little bit of hell. The ride will take about an hour and 40 minutes—assuming average traffic—and set you back $110 or more (not including tip). The good news is that Uber plans to fix that—slashing the time to 15 minutes and the cost to $20. The wrinkle is that the way the company plans to achieve that breakthrough is with a technology that has been one of the modern world’s great—and most elusive—dreams: the flying car.

Uber made that big announcement back in October, when it unveiled Uber Elevate, an ambitious program to get cars—or at least Ubers—off the road and into the sky. Yesterday the company went a step further, announcing that it had hired engineer Mark Moore, a 30-year veteran of NASA, to develop the technology. With that, the easy part ended.

Some of the challenges facing the Uber flying car plan are the same ones that have plagued every such plan ever proposed: safety and congestion. Filling the skies over major urban centers with cars is an order of magnitude more dangerous than filling the streets with traditional cars, since at least an accident on the ground stays on the ground. An accident in the sky threatens everybody below.

Uber is hardly unmindful of this matter and devotes a section in its Uber Elevate proposal to the problem of air traffic control. The closest current analogy to what Uber envisions would be the air taxi helicopters that operate in some urban centers. New York City, for example, has a service that offers “quality airport shuttle services” in and out of Manhattan, but at $1,750 per booking (again, not counting tip). That keeps the client base small and the congestion low.

A better analogy, Uber argues, is São Paulo, Brazil, which has hundreds of helicopters in its commuting fleet and boasts a good safety record all the same. But Uber itself admits that a “successful, optimized, on-demand” air system “will necessitate a significantly higher frequency and airspace density of vehicles operating over metropolitan areas.” Handling that flying crush, Uber says, will require the development of entirely new types of air traffic control systems.

Certainly, that’s not an insurmountable hurdle, and solving such problems is the reason America’s engineering institutes—academic, commercial and governmental—get out of bed in the morning. So if new air traffic control systems are needed to make air taxis a reality, the betting is that they will be.

The bigger—much bigger—challenge will be the vehicles themselves. Small airplanes are obviously out of the question for traveling into and out of urban centers, since runways are one thing downtowns don’t have. Helicopters will always be a boutique product, not just because they’re expensive but because they’re loud and heavily polluting. The answer is the vertical-takeoff-and-landing (VTOL) vehicle, an aircraft that takes off and lands like a helicopter but flies horizontally like an airplane.

Uber made a smart choice in selecting Moore to head its R&D initiative. He has worked extensively in VTOL design for NASA and in 2010 released a surprisingly readable white paper describing his ideas. Both he and Uber envision electric-powered craft—a must if cost, noise and emissions are going to be kept under control—and he has studied high energy-density battery development as well.

The problem is, the new proposals are not remotely the first VTOL plans that have ever been floated. The technology has a long history—not much of it terribly good. The greatest of the cautionary tales is the V-22 Osprey, an American military vehicle first proposed in 1980, after the Carter Administration’s disastrous attempt to rescue the 53 American hostages held in Tehran. The mission failed due to mechanical breakdowns in three of the helicopters dispatched for the rescue and a crash of the fourth one. That made clear the need for a whole new kind of aircraft capable of a stealth and nimbleness helicopters aren’t, but things turned out not to be so easy.

As TIME reported in a 2007 cover story, the Osprey fleet, which was supposed to be operational by 1990 at a cost of about $2.5 billion, did not deploy into a war zone until 2007, by which point the overall price tag for the program was projected to be $55 billion. The biggest challenge was the vehicle’s tilt-wing design—with propellors on pivoting wings that point up during takeoff and landing and forward during flight. Not only did the wing technology itself prove difficult, but the aerodynamics of prop wash suddenly changing directions could be exceedingly tricky. Thirty people died in crashes during the plane’s development and while Ospreys have been used in combat support and humanitarian roles around the world, they never achieved anything like their promised potential.

That hasn’t kept the military from trying. In 2013, an upgraded Osprey was made available as part of President Obama’s White House fleet, but while reporters, support staff and even first dog Bo were cleared to fly aboard the thing, the President himself was forbidden due to safety concerns. Currently, DARPA—the Defense Advanced Research Projects Agency—is working on yet another VTOL iteration, hoping this time it will click.

It is undeniably a good thing that government designers keep pressing ahead with VTOL technology. It is in some ways an even better thing that Uber has announced its own initiative, since there’s nothing like a monetizable product or service to get the inventive energy going. But there’s a reason flying cars have been a world-of-tomorrow fantasy for so long. Promising them is one thing; producing them is another. Uber may well meet that challenge, but it’s not likely to be easy.

This story originally appeared on Fortune.com. Copyright 2017