There's a reason Google is testing its autonomous vehicle prototypes in sunny California.

The heavy snows that have blanketed much of the country in recent weeks might have you wishing for the day when you can buy an autonomous vehicle. Imagine not having to squint through a frozen windshield as you crawl through traffic on your way to work.

That’s one of the big promises of self-driving cars which, proponents promise, will make driving easier and safer. The first semi-autonomous vehicles are set to reach showrooms within the next several years, and several automakers are promising they’ll have the first fully-autonomous products on the road by 2020.

That timetable might work in some parts of the country. But if you live in snowier climes, don’t be surprised if you’re in for a much longer wait, especially if you’re hoping to have your car take over driving duties in inclement weather.

Google recently began testing what will soon be a fleet of at least 100 self-driving prototypes near its Silicon Valley headquarters. Most of those pilot vehicles won’t even be equipped with steering wheels and pedals, just a microphone where passengers can input a destination.

“The Google car is a good show car, but it’s not going to be something you can drive on (Detroit’s) I-75 when it’s snowing,” cautions Steffen Linkenbach, director of the Systems and Technologies group for automotive supplier Continental North America. “It will be unacceptable to customers that the system won’t work because it’s snowing or the sun is shining at the wrong angle,” he adds.

(New vehicle designs make goal of zero highway deaths a real possibility, says new report. Click Here for more.)

One of the largest suppliers of key driver assistance components, Continental has become one of the leaders in autonomous vehicle research. But it is pushing back on the aggressive timetables some of its automaker clients have laid out for putting self-driving technology into production.

And it’s not alone in suggesting it will take longer than promised to develop cars that can operate autonomously anytime, anyplace.

(American motorists are ready for autonomous vehicles and willing to pay for them. Click Here to learn more.)

“I’d give anything to be able to turn over control during that half hour commute,” says Sam Abuelsamid, a senior analyst with Navigant Research.

But he notes that even the more limited driver assist technologies showing up in the latest luxury models – such as the tech-laden Mercedes-Benz S-Class – can’t always deal with all the conditions a vehicle must operate under. The traction control systems found on all vehicles now sold in the U.S. can’t always distinguish between icy roads, gravel and fresh snow, for example, “but how you want to have a car respond on each surface is very different,” Abuelsamid points out.

The most basic driver assistance systems rely on wheel and yaw sensors to get a sense of what’s happening to a vehicle. More advanced cars now add a blend of radar, sonar, cameras and an advanced laser technology known as LIDAR.

Bringing them all together is an increasingly complex task. Even today’s more advanced vehicles, like the S-Class, use dozens of times more software code than an F-16 fighter jet. Fully autonomous systems will bump that up by several orders of magnitude.

And that’s just to attempt to replicate what a human driver does every day, says Jim Sayer, an autonomous and connected vehicle research scientist at the University of Michigan Transportation Research Institute.  “We seldom think about just how complex the act of driving is, and how much information we have to take in as we proceed down a roadway.”

While driver error is blamed for 90% of all collisions, humans still can do a much better job at estimating where to point their vehicles on a snow-covered road, where sensor technologies currently have to give up because they can’t make out lane markers, for one thing. They also can be fooled by streaks of salt left on dry pavement from the last storm, and even by bright sun glaring off the road at the wrong angle.

Complicate matters by having snow, ice and salt grime covering those sensors.

Sayer, Abuelsamid and Linkenbach all agree that industry researchers will eventually come up with solutions to these problems. But until they do, autonomous vehicles won’t be ready for prime time, at least not in those challenging environments where they might be most desirable.

(Winter storms couldn’t slow down January car sales. Click Here for the story.)