Forget the motor and the drivetrain. The main engine of the self-driving car of the future will be an AI-powered supercomputer. That future could be approaching faster than you think: Next month Nvidia will release a self-driving car computer, one of the first to hit the market. Called Drive PX, it has 2.3 teraflops of processing horsepower, 12 camera inputs, and computer vision algorithms—essentially it’s an automotive OS that automakers and developers can use to help “teach” cars how to drive.
Making smart decisions quickly is one of the big promises of artificial intelligence. This is why many see automated driving systems as the future of transportation. Of course, it’s a leap of faith to turn control over to robots, yet most airplanes cover the majority of their routes on autopilot, so it isn’t without precedent. (And the recent Lufthansa tragedy may open the door to pilotless air travel sooner than we think.)
Today, many cars already pump the brakes when sensing danger or stay in-lane if a driver takes his or her hands off the wheel. Lexus has had a self-parking car since 2007. Tesla Motors’ Model S “autopilot” will soon feature push-button lane changes. And legacy brands like BMW and Mercedes are pushing these changes, too. But as in the case of longtime game and entertainment-tech maker Nvidia, next-gen innovation is coming from a variety of technology ventures, big and small—not Detroit. Here are a few worth a second look:
• Automatic— Its diagnostic module plugs into your car’s computer to analyze driver habits, trip costs, fuel efficiency, and other hidden performance metrics.
• Google X—The search giant’s well-known innovation lab has been training cars to drive on their own by learning roads, signals, signs, and how to detect danger.
• Seeing Machines—In-vehicle cameras that monitor a driver’s eye movements to alert them when they’re distracted, drowsy, or drunk.
• IBM—Big Blue has been using its other-worldly powers to test traffic optimization in big cities around the globe.
• Navdy—Head-Up Display (HUD) that projects map displays as transparent images floating outside a car’s windshield, letting drivers keep their eyes on the road.
To be sure, companies and legislators will have to answer some tough questions before humans hand over the keys. Who’s at fault if two self-driving cars collide, for instance? How can we prevent car hacking so that cars don’t route us toward danger? What if people actually want to drive? Elon Musk suggested recently that lawmakers may eventually “outlaw [human-] driven cars because they’re too dangerous.” One thing is for certain: As the technology advances, so too, will our definition of what it means to take the wheel.
No comments:
Post a Comment