Tesla's autopilot lets cars drive, change lanes themselves

October 14, 2015 byDee-Ann Durbin
Tesla's autopilot lets cars drive, change lanes themselves
In this Sept. 29, 2015 file photo, Elon Musk, CEO of Tesla Motors Inc., introduces the Model X car at the company's headquarters in Fremont, Calif. Tesla Motors on Wednesday, Oct. 14, 2015 unveiled an autopilot system that lets its cars change lanes by themselves. It will be added to some Model S sedans and Model X SUVs through a software update. (AP Photo/Marcio Jose Sanchez, File)

Electric car maker Tesla Motors is leapfrogging competitors with a new autopilot system that lets cars change lanes by themselves.

Like other semi-autonomous systems already available from Mercedes, Audi and Volvo, Tesla's system automatically keeps the car within its lane and maintains a certain distance from the car in front, both at highway speeds and on city streets. It can find a parking spot and parallel park itself. It also uses cameras and sensors to warn drivers about potential side impacts.

But analysts say the lane-changing feature is an industry first. Tesla CEO Elon Musk said the system is also unique because it will constantly collect data from actual drivers and improve itself. The system will note, for example, how quickly drivers can safely navigate a particular bend in the road or where stop signs are located.

"I think this is going to be quite a profound experience for people," Musk said Wednesday in a conference call with media. "It will change people's perception of the future quite drastically."

Musk also added a word of caution: Drivers need to keep their hands on the wheel, and the autopilot system will chime to remind them if they don't. Drivers—not Tesla—will be held liable if there's a crash, Musk said.

"We're being especially cautious at this early stage, so we're advising drivers to keep their hands on the wheel just in case," he said. "The software is very new."

Tesla's autopilot lets cars drive, change lanes themselves
In this Sept. 15, 2015 file photo, a Tesla Model S is on display on the first press day of the Frankfurt Auto Show IAA in Frankfurt, Germany. Tesla Motors on Wednesday, Oct. 14, 2015 unveiled an autopilot system that lets its cars change lanes by themselves. It will be added to some Model S sedans and Model X SUVs through a software update. (AP Photo/Michael Probst, File)

Musk said fully autonomous, hands-free driving is still at least three years away from a technical standpoint, although it will probably take regulators longer than that to allow it.

The autopilot update will be added to around 60,000 vehicles worldwide, including Model S sedans made after September 2014 and Model X SUVs. Owners will get the system through a software update starting Wednesday evening in North America. Owners in Europe and Asia will get the software update in about a week. People with Model S sedans that were made earlier don't have the required sensors and won't be able to add them retroactively, Musk said.

Only owners who paid the $2,500 charge for the full autopilot system will be able to activate all of the autopilot features, but Musk said the side-impact warning is a safety feature and will be available to everyone. For the next update, Musk said, Tesla is working on having the car drive itself in and out of garages when it's summoned by the owner.

Karl Brauer, a senior analyst with Kelley Blue Book, said Tesla's system appears to up the ante for the industry. General Motors' Super Cruise system, for example, which is due out on the 2017 Cadillac CT6, will let drivers take their hands off the wheel at any speed on the highway, but it won't change lanes by itself.

"This is the game we're going to be playing, round and round, for the next five to ten years until there's fully autonomous driving," Brauer said. "Each time there's a step by someone out there, everyone will have to match it."

Brauer said there's a danger to that, because people may not know what the car they're driving is capable of.

"Having that kind of nebulous, blurry gray zone could introduce many more problems," he said.

But Musk is optimistic about the benefits of autopilot.

"In the long term, it will be safer than a person driving," he said.

Explore further: Tesla updating Model S to ease range anxiety, improve safety

Related Stories

Recommended for you

Microsoft aims at Apple with high-end PCs, 3D software

October 26, 2016

Microsoft launched a new consumer offensive Wednesday, unveiling a high-end computer that challenges the Apple iMac along with an updated Windows operating system that showcases three-dimensional content and "mixed reality."

Making it easier to collaborate on code

October 26, 2016

Git is an open-source system with a polarizing reputation among programmers. It's a powerful tool to help developers track changes to code, but many view it as prohibitively difficult to use.

Dutch unveil giant vacuum to clean outside air

October 25, 2016

Dutch inventors Tuesday unveiled what they called the world's first giant outside air vacuum cleaner—a large purifying system intended to filter out toxic tiny particles from the atmosphere surrounding the machine.


Adjust slider to filter visible comments by rank

Display comments: newest first

1.7 / 5 (6) Oct 14, 2015
What the hell ? This function can be coded in 15 minutes by an engineer with sufficiently high level methods for detecting objects.
If competitors didn't put this function online, it's because it has greater probability for accidents than single lane driving. No revolution here, just Tesla taking a risk.
not rated yet Oct 15, 2015
with sufficiently high level methods for detecting objects.

That's the issue.

There isn't really a practical, cheap, and dependable radar or vision system for cars that could do that with the limited processing power and time available.

One can compensate for the extremely limited perception with more intelligence to build a more complete picture for the algoritm to work on, but since such object recognition from limited cues requires a level of understanding of what you're looking at, and all the AIs are actually dumb as rocks in this respect because they rely on simple statistical analysis based on a training set (ie. "googling" a huge database of pictures), the whole thing is basically just a liability lawsuit waiting to happen.

They know it will mess up. Hence why, "hands on the wheel".

1 / 5 (2) Oct 15, 2015
It's easy to dismiss how much processing a human driver actually goes through.

All you need is one glance at the side view mirror, and a half-a-glance at the dead spot to determine that it's empty, the road behind is clear and you can switch lanes.

The computer would choke simply by the view of the mirror, because the distortion of the glass throws off its distance estimates.

If you think human drivers are blind idiots, the computer is basically like Terry Schiavo. That's why Google relies on a sweeping laser radar on the roof to have a reliable view of any big solid objects around. The laser radar however isn't cheap, it's limited in what it actually sees, and it isn't robust or reliable in the mechanical sense because a single bug splat will disable the car's vision completely.

1 / 5 (2) Oct 15, 2015
We could all be driving cars with the following characteristics:

For no reason whatsoever, your car would crash twice a day.

Every time they repainted the lines in the road, you would have to buy a new car.

Occasionally your car would die on the freeway for no reason. You would have to pull over to the side of the road, shut off the car, and restart it, before you could continue. For some reason you would simply accept this.

Occasionally, your car will shut down and refuse to restart, in which case you would have to reinstall the engine.

The oil, water temperature, and alternator warning lights would all be replaced by an irrecoverable exception error.

Anti-Virus software will be mandatory and expensive.

1 / 5 (2) Oct 15, 2015
Here's what the Google car actually sees of its surroundings:


The picture is a bit blurry because it's a photograph of a screen, but you can see all the other cars around the Google car in the traffic lights being just vague brown blobs, and the algoritm has drawn a box around them to say "I think this might be a car". There's also several blobs that aren't boxed - maybe they're just glitches, maybe they're motorcyclists - who knows? Certainly the car doesn't.

It's unsettling how little information the car actually has of its surroundings and what little it makes of it. All the good sharp lines, where the lanes are, the traffic lights etc. come from a pre-programmed map of the area, which the computer tries to align to its actual position by guessing where the curbside is from the lidar image.

If the map is wrong, the car has no idea.

Please sign in to add a comment. Registration is free, and takes less than a minute. Read more

Click here to reset your password.
Sign in to get notified via email when new comments are made.