The reason, Musk told The Journal, is that Tesla — and others — are still trying to crack the code of helping computers recognize objects. Current systems rely on radar, cameras, and other sensors to see what’s around them and make decisions. Some of those systems have gotten smaller, better, and less expensive, though they still require software that can identify objects and make the right decisions.
CARS ARE STILL LEARNING HOW TO RECOGNIZE THINGS
Many other companies are currently trying to perfect just that process, including Google which is running virtual simulations of California roads to train the cars used in its self-driving car project. For its part, California this week issued the first group of permits to let Google, Mercedes, and Audi legally begin testing self-driving cars on roadways, with other companies expected to follow.
The two major promises of self-driving cars are safety and convenience. Computer-controlled cars promise to react to things faster, and could open up certain sections of roadways to higher speeds given the extra reaction time — speeding up long distance car travel. “They will be a factor of 10 safer than a person [at the wheel] in a six-year time frame,” Musk says. For convenience, a car that drives itself would allow passengers to focus on other things besides maneuvering around roadways and other drivers.
Musk’s estimates for Tesla are not that far off from other automakers, which hope to have fully-autonomous cars on the road by 2020. Last August, Nissan said it planned to have multiple modelsavailable by then, with an estimated price increase only of $1,000 per car. Others, like GM andstartup Cruise are aiming to deliver souped up versions of cruise control that will let the car drive itself on highways, just with a human still behind the wheel.