NASA engineers referred to the most crucial moments in the Mars Curiosity mission last summer as the "seven minutes of terror." During that time, without the guidance or control of any humans back on earth, the rover had to enter the martian atmosphere, descend from an initial speed of 13,000 miles per hour, and land on target in one insanely expensive, intact piece.

"And the computer," as the engineer in this dramatic little NASA trailer explains, "has to do it all by itself":

Perhaps this sounds like a leap too far in transportation technology, but if engineers can build a car-sized robot capable of managing its own internal errors in a first foray to the martian landscape, surely some less sophisticated but equally reliable systems could be built into our more terrestrial vehicles?

"Imagine the complexity in software and from the hardware perspective that’s necessary in order to launch that successfully in seven minutes," says Aziz Makkiya, a design telematics engineer at Ford. "That’s one of the experiences that we're trying to mimic in a smaller scale in the automotive world."

Cars are about to get substantially more complex, more reliant on computers. Soon enough, they'll automatically be talking to each other, to the infrastructure around them, to distant emergency responders. And this isn't just in the faraway world of driverless cars. Cars that still have people behind the wheel will have "connected vehicle" technology in them that simply makes them safer, by for instance registering the presence of nearby speeding cars.

All of this technological promise, though, comes with greater risk. Parts breaks. Networks fail. Sometimes our phone calls get dropped or the power goes out. The real question about connected cars is actually the same one NASA asks about all those robots in space the agency invests millions of dollars in: How do we make sure this stuff never fails?

"It’s capable of handling internal errors," Aziz says of the types of robotics that are used in space. "That’s what we’re trying to accomplish."

Ford has just announced a three-year research project with the St. Petersburg Polytechnic University in Russia to study how robots communicate in space with the goal of figuring out how cars might best communicate on earth. In space, robots communicate with each other, and with astronauts in the International Space Station, and with humans monitoring all of them back on earth. Connected vehicle technology faces a similar challenge in needing to direct information with varying priorities, through multiple channels—all with backup systems to the backup system to the backup. A Ford engineer describes the project here:

Ford is aiming to develop one platform with multiple communications technologies on it—dedicated short-range communication, cellular wireless broadband, mesh networks—that would include a kind of smart switch capable of automatically toggling between the networks depending on the situation.

Say, for example, that your Prius is in imminent danger of ramming into another car ahead of you. "This is a high-priority message," Aziz says. "You would like to transmit that as fast as possible, and in that case DSRC or a mesh network is the right medium." But maybe your car needs to communicate with a toll authority, or a rental company. "Then we would like to fall back to different technologies if it saves the bandwidth and if it saves the cost."

The trick is to figure out, automatically, which networks are best in which situations, while traveling at 50 miles an hour, and without ever losing connectivity. "Think of it as a soft hand-off between one technology and another," Aziz says.

Part of the goal, obviously, is to pull all of this off at a cost that would make the technology feasible in millions of automobiles. NASA probably isn't the best model on how to scale up affordable mass-market technology. But this also wouldn't be the first technology to start, in theory, in space before trickling down to the rest of us.