Learning Center

// -->

Future Tech: Self-Driving Cars

Talk about self driving cars is everywhere. Only yesterday I was reading about how one self driving car operated by Google cut off a second self driving car operated by a competitor whilst they were both on the road. The world’s very first instance of computer-controlled road rage incidents, although to be fair to the self driving cars, they kept silent and drove on.

Google is probably the most high profile company working on developing the technologies that would allow vehicles to drive themselves autonomously, but secretly so is everybody else too.

Uber wants taxis that collect you without drivers; Mercedes wants trucks that can deliver goods across the world all night without stopping; and in the agricultural world John Deere has already turned tractors and farm machines everywhere into self-driving versions, making farmers very happy.

I think it’s a safe bet to say that self driving cars are going to be with us commercially by 2020. I think the technology will get there before the legislation does though.

Based on the many articles that I’ve read on the subject, I somewhat doubt that a fully-autonomous car that’s capable of driving anywhere under any conditions will be ready until at least 2020.

To get an idea of where we are at now, though, let’s take a look at all of the major autonomous car projects currently in progress and check out what they’ve accomplished so far and what their goals for the future are.

Google is generally thought to be the furthest along of any of the self-driving car developers. They are also the most high profile in the space.

It began its project in 2009, and its cars have already self-driven more than 700,000 miles, according to the company. For its tests, Google has mostly used modified versions of normal, commercially-available cars, including the Toyota Prius and Lexus RX450h, though in 2014 it also designed and had manufactured its own self-driving car models, including some without any steering wheels or pedals at all.

Each prototype reportedly has over $150,000 in specialized equipment onboard, including a $70,000 LIDAR system to develop a real time 3D map of the car’s surroundings.

Limitations of Google’s self-driving car technology include the inability to work when it’s raining or snowing heavily, to identify and avoid potholes, and to know when an object, such as an empty trash bag, is safer to run over than avoid. That trash bag could be someone's dog, so these small things are important to get right before the technology goes live.

Perhaps most disappointingly, it also appears that at the moment Google’s self-driving cars can only be driven in areas that Google has digitally mapped with its Google Streetview cars, although admittedly that’s pretty much everywhere you are likely to want to (self) drive your car anyway.

According to Lee Gomes of the MIT Technology Review, the company’s cars have only been tested where “intricate preparations have been made beforehand, with the car’s exact route, including driveways, extensively mapped.” So the technology, although impressive is still not quite there.

Whether Google will manufacture and release its own self-driving cars (think crazy colored bug cars speeding you around from A to B) or sell its self-driving technologies to car manufacturers remains to be seen.

Google has reportedly already been discussing such an arrangement with companies such as Volkswagen, Daimler, Ford, GM and Toyota, but we also know that many of these companies are also trying to develop self driving capabilities of their own.

The car manufacturers that have made the most progress in developing self-driving cars themselves are Nissan, Audi, Volvo, Daimler and Tesla.

Nissan so far is pretty much the only auto company that has matched Google in demoing a self-driving car on public, city streets (as opposed to highways) to journalists, in video camera-equipped Nissan Leafs.

Nissan’s test drives, like Google’s, had limitations. They were short drives and took place in a (presumably) low-traffic industrial area.

In addition, Nissan’s self-driven car also partly relies on previously-assembled mapping data to orient itself, though the company appears to be trying to limit the car’s dependence on this data: according to McCracken, “Nissan…uses what it calls sparse maps. They’re less detailed [than Google’s], and based on third-party data. ‘Part of the research is to figure out how sparse the sparse map can be, and what information needs to be in it,’ said [director of the Nissan Research Center Silicon Valley Maarten] Sierhuis.”

So again the stumbling block is that you need to have at least some of the road digitally mapped before you can send a self driving car down it.

It should be noted, though, that Nissan has never formally announced that it would build a fully-autonomous car; rather Nissan’s goal is to introduce semi-autonomous features that will handle “most, but not all, of the heavy lifting of driving” into Nissan, Infiniti and Renault vehicles by 2020.

Audi and Volvo, for their part, have both demoed self-driving cars on public highways for journalists, although this is not quite as impressive and even more of a controlled test than those Nissan’s or Google’s.

In Audi’s demonstration, it took several journalists in modified, self-driving Audi A7s (with “six radars, three cameras, and two light detection and LIDAR units”) from northern California to Las Vegas—a trip of 560 miles.

Impressively, these cars were able to drive 70 mph and switch between lanes by themselves. Limitations include the inability to drive at night, in poor weather, and on anything but highways with well-marked lanes.

The Audi team had also “run the route” at least six times beforehand. Volvo’s self-driving car demo in a modified S60 sedan, meanwhile, was limited to only about 35 miles of highway in Gothenburg, Sweden.

Most ambitiously, Volvo has set a goal of reducing the number of deaths and serious injuries to its car owners to zero by 2020, with autonomous and semi-autonomous driving features having a key role in this effort.

Despite the greater speeds of highways, city or surface streets are actually more difficult for self-driving cars to handle due the increased complexity of the environment—with tricky elements such as pedestrians, dogs, cyclists, traffic lights, turns, four-way stops and two-way roads without marked lanes.

Many higher-end cars essentially already have the ability to drive themselves on well-marked freeways with semi-autonomous features like adaptive cruise control and automatic lane keeping, though these features are really only designed to occasionally assist or correct human drivers rather than replace them, and such cars usually require drivers still.

Tesla has incorporated many semi-autonomous features into its Model S, and is planning to release a software update that would allow Model S vehicles built since fall 2014 to drive themselves on highways, perform safe lane changes automatically with a flick of the lane change signal, and park and summon itself when on private property. Imagine that!

Despite the advanced semi-autonomous features in their cars, however, Tesla has yet to demonstrate a self-driving car on public roads in the way that Google, Nissan, Volvo, Audi and Volvo have.

As for BMW, they have mostly dismissed the idea of building fully-autonomous cars, arguing, reasonably enough, that the only reason people buy BMWs is because they enjoy the experience of driving them.

BMW’s opposition to self-driven cars brings up an interesting point, which is that there are barriers to the adoption of autonomous vehicles besides just the limitations of current technology: cultural and legal barriers.

So depending on how much or little self driving autonomy you are looking for, you should have something to look forward to in the next five years, even if the latest technology will cost a small fortune and be in the most expensive cars. Being an early adopter is about monetary sacrifice mostly.

How autonomous you want to go will depend largely on where you are and how well the roads have been digitally mapped out beforehand, so bear that in mind when you ask your Google car to take you down to Baja California for a weekend.