Source: NVIDIA.
Much of the talk surrounding autonomous vehicles make it sound like they'll be nearly ubiquitous in the next few model years. Tesla (TSLA -3.55%), Alphabet's (GOOG 0.37%) (GOOGL 0.35%) Google, NVIDIA (NVDA 0.76%), and Delphi (DLPH) are certainly all moving in this direction. But the reality of self-driving cars flooding the roads is likely further away than the companies, and consumers, would like.
To understand why, we need to take a look at where autonomous car technology is succeeding, and what obstacles are pushing against it.
Why we're on the path to autonomous cars
We are quickly heading down the road to self-driving cars, there's no doubt about it. Google has already poured an estimated $60 million into its autonomous vehicle research and testing, according to IHS Automotive.
Source: Google.
The company's been testing self-driving vehicles since 2009, and its vehicles have already racked up 1 million miles of autonomous driving. The company has two types of autonomous vehicles right now, including a fleet of Lexus SUVs Google's been using since 2012, and the new self-driving cars that the company's built from the ground up. Both types of vehicles have been awarded self-driving permits in California, Texas, and Washington.
If there's another company that talks nearly as much about self-driving cars as Google, it's probably Tesla. The electric carmaker already has an Autopilot feature on its Model S that can steer the car, make lane changes, and automatically adjust speed to match the traffic flow.
A new version of the car's software also allows Tesla owners to summon their vehicles to come to them, and to park themselves without the driver inside the car.
But Tesla CEO and co-founder Elon Musk wants his cars to go further. On a conference call back in December he said that he envisions fully autonomous Teslas on the road two or three years from now. He said he envisions a Tesla owner living in New York summoning their car to drive itself to them all the way from Los Angeles.
And then there's NVIDIA, which is known for its mobile graphics processors, but is emerging as an autonomous car player because of its Drive PX platform. The software is paired with a powerful computer -- mounted in the trunk of the car -- that processes information in real-time to evaluate the car's surroundings and the distance from nearby cars, identify pedestrians, and control the vehicle.
At the 2015 Consumer Electronics Show (CES) the company teamed up with Audi to have an A8 sedan drive itself from San Francisco to Las Vegas using its system. And at the 2016 CES, NVIDIA introduced the latest version of its autonomous system, the Drive PX 2 Platform.
Delphi is taking a similar approach to autonomous vehicles. The auto parts supplier makes self driving hardware, and used its technology to send an Audi from San Francisco to New York (across 15 states and about 3,400 miles), nearly all on its own, last year. Delphi said the car drove 99% by itself (mainly on the highway), but needed a human to drive on city streets.
Source: Delphi.
If the brief examples above weren't enough to prove that autonomous vehicles are well on their way, the U.S. government recently pledged nearly $4 billion over the next 10 years to help the government work with carmakers and tech companies to establish a clear path to autonomous vehicles.
This is all very promising for the autonomous vehicle segment, but there are a few problems that could stifle autonomous vehicle progress in the near future.
Pumping the brakes on autonomous tech
Despite the proposed autonomous vehicle funding, federal and local governments could still be major inhibitors of self-driving technology progress.
The Department of Transportation said recently that it will release automated driving standards for states to use as a model in the next six months. That's a big step forward over the preliminary standards set up a few years ago, but it's still anything but certain.
States establish their own autonomous driving rules right now, and, unsurprisingly, only a handful of states even allow real world testing of these vehicles. The upcoming standards could help change that, but they'll only act as a model for the states and won't be hard and fast laws.
Source: Tesla.
But let's pretend for a moment that the federal government and state governments all come together in a Utopian fashion and self-driving standards are adopted quickly across the U.S. Even in that scenario, there's still a lot of liability issues that will be raised.
As more technologies are integrated into autonomous cars, the responsibility each company has if something goes wrong becomes much murkier. Is a driver responsible if a car hits a pedestrian while driving autonomously? Or does the carmaker, the software maker, sensor maker, map service, etc. take the blame?
Volvo made the news a few months ago when it became the first automaker to take full responsibility for autonomous vehicles when they're in autopilot mode. Others could follow suit, of course.
But right now, it's fairly easy for Volvo to take full responsibility for its autonomous vehicle systems because the company hasn't sold a single autonomous vehicle to a consumer. Automakers can make all the claims they want right now because we don't yet live in an autonomous car reality. The real test for automakers will come when a self-driving car makes a wrong decision (yes, it will happen) and causes a major accident. Only then will we know if automakers will be ready to really shoulder the blame.
And then there's security. Many people may be unconcerned about their smartphone or computer's security, but the same won't be true for cars. Handing over complete control to an autonomous system will already take a great deal of trust, even more so knowing that someone could maliciously hack the vehicle and take over the system.
In case you missed all the reports from a few months back, hackers successfully tapped into a Jeep Cherokee and were able to remotely control the braking, acceleration, steering, and a host of other controls while a Wired journalist sat behind the wheel. And this wasn't even an autonomous car. As a result, Fiat Chrysler recalled 1.4 million cars.
Carmakers will soon carry the burden of not just ensuring autonomous cars are fit to drive themselves, but that they're also 100% secure, all of the time. This too, will take time to develop and perfect.
Source: Ford.
Lastly, we're still at the beginning stages of testing fully autonomous vehicles in real world scenarios. Just last month, Ford become the first automaker to test its self-driving cars in winter weather. That's a great step toward full autonomy, but it also means we're still far away from having autonomous cars driving us around in every scenario possible. Ford says that nearly 70% of the U.S. population lives in snowy regions. If carmakers are going to stand behind their autonomous systems even in the worst weather, it's likely they'll need years of testing.
When we reach self-driving realization
IHS Automotive expects 12 million self-driving cars to be sold annually by 2035 and autonomous car ubiquity "sometime after 2050." Those are just predictions, of course, but I think, considering the hurdles above, that's a realistic timeframe. Which means that while there's a lot of hype around the market right now, it's likely that we're still decades away from a time when the masses start getting behind the wheel -- er, dashboard -- of a fully autonomous vehicle.
