The Future Of Self-Driving Cars

We could start seeing a lot more of these futuristic cars on the road soon.

(You’ll find the text version of “The Future Of Self-Driving Cars” below this embedded video)

Self-driving cars like these were supposed to be a common sight by now, but the road to autonomy has turned out to be a lot bumpier than we thought.

Let’s talk about what the hold up has been and why we could start seeing a lot more of these futuristic cars on the road soon.

In 2016, Business Insider predicted that there would be 10 million driverless cars on the road by 2020. A year later, the CEO of Ford predicted that his company would manufacture an autonomous car with no gas pedal or steering wheel by 2021. Companies like General Motors and BMW made similar claims.   

But more than five years have passed and driverless cars are still a rare sight.

The reason those optimistic predictions from several years ago turned out to be wrong has to do with what you could call “the challenge of the nines.”

When those predictions were made, people in the industry believed that driverless cars were on the cusp of being safe and reliable enough to operate on public roads.

There was this idea that driverless technology was 99% of the way towards being ready for widespread adoption, and it just needed a few more improvements to get there.

99% was a huge achievement, but it still wasn’t good enough. No one wants to get into an accident once every hundred times they go for a drive.

Perfect might not be attainable, but the goal was to get as close to 100% reliability as possible—something like 99.99% or more.  

That seemed doable, even easy. A less than 1% improvement in reliability didn’t seem like much.

But it turns out that going from 99% reliability to 99.99% reliability is extremely difficult.

Those extra nines represent edge cases— situations that drivers encounter rarely, like a pedestrian suddenly running into the road or an unexpected detour due to construction.

And because those types of scenarios so rare, self-driving cars haven’t had access to as much real-world data to train their AI algorithms on. Instead, they’ve had to rely on simulations to try and anticipate and prepare for those situations.

That’s what self-driving car companies like Waymo, Cruise, and Zoox have been working on all these years. They’ve been improving the reliability of their cars so that they can meet the high safety standards that are required for widespread adoption—a process that’s been a major slog.

Progress

But still, even though progress on autonomous vehicles has been slower than expected, there has been progress—and we could be getting close to the point where you’re going to see many more of these cars on the roads soon.

Today, anyone in Phoenix can hail a completely autonomous vehicle powered by Waymo’s technology to drive them around. The company also offers limited rides in San Francisco.

Meanwhile, Cruise, a competing company, recently launched driverless ride hailing service in San Francisco and more limited services in Phoenix and Austin.

Another company, Motional, announced that it would soon offer driverless rides in Los Angeles and Las Vegas.

As you can see, completely self-driving cars are available to the public today in a growing number of cities in the U.S.—something that wasn’t the case a few years ago (Baidu and Pony.ai plan to start offering self-driving rides without backup human drivers in Beijing, China this year).

These cars are being introduced to the world through robotaxi services. They’re like Uber but without the human driver.

Robotaxis are going to be the first way that people experience autonomous transportation.

The day when you or I can go out and purchase a fully self-driving vehicle for ourselves is probably still a ways off, contrary to what Tesla CEO Elon Musk has been saying.

Competing Visions

Companies like Waymo and Cruise build their cars to be autonomous from the start. That’s a different path than the one laid out by Tesla, another company that has ambitions to develop autonomous vehicles.

Tesla’s goal has been to slowly turn its electric vehicles into autonomous vehicles by incrementally upgrading them with new self-driving software.

In 2020, Tesla unveiled its initial Full Self Driving software update, but it’s been a disappointment for many. The Tesla cars equipped with the new software aren’t anywhere close to autonomous.

What Tesla calls “Full Self Driving” is more like an advanced driver assistance system. This would make Teslas today Level 2 vehicles under the five-level driving automation scale created by the Society of Automotive Engineers.

This is far from the level 4 or the level 5 that would need to be achieved before a car can drive without any human oversight— something that Waymo and Cruise cars are already capable of.

A few months ago, Tesla was sued by a group of customers who claimed that the company misled the public about its cars’ self-driving capabilities. These customers are upset that they paid thousands of dollars for an upgrade that doesn’t do what it’s supposed to do.

For years, Elon Musk has said that Tesla cars will be capable of full autonomy using their existing hardware. He promised that in 2020, over a million Teslas would be capable of fully autonomous driving. Yet today, there isn’t a single one capable of anything close to that.  

It remains to be seen whether Tesla’s path to self-driving cars is even viable. The company has taken a very different approach to autonomy than others like Waymo and Cruise.

Teslas today lack some of the sophisticated hardware that Waymo and Cruise cars have, like Lidar sensors, which uses lasers to measure distance and create a 3D map of the surrounding environment.

Tesla believes that full autonomy can be achieved with computer vision systems using cameras alone and has built its self-driving technology around this approach.

But some people, like Waymo’s former CEO John F. Krafcik, have said that Tesla’s approach to autonomy is a dead end that simply won’t work.

We’ll see who is right. Today, Waymo and Cruise have the upper hand. They actually have driverless vehicles on the roads, something that Tesla seems far away from achieving.

Expansion

Not only do Waymo and Cruise have driverless cars on the road, they are also slowly expanding the number of cities in which they operate.

It’s amazing to me that you can be in Phoenix and San Francisco and actually get a ride in a driverless car. It’s like something out of science fiction. And now that we’ve seen their success, there’s little doubt that these companies can continue to expand into cities with similar driving conditions.

After that, the question becomes: can these vehicles operate in more difficult environments, such as areas with steep cliffs or heavy snowfall?

That’s something that remains to be seen. A fully autonomous level 5 vehicle that can operate in any environment without human intervention is something that is still a ways off, and something that some people believe may never come to fruition.

But that’s probably too pessimistic. We already have legitimate level 4 autonomous cars that can operate without human drivers in specific environments. Maybe that’s all we’ll have over the next several years, but technology is improving at such a rapid rate that it would be foolish to rule anything out.  

We also have to take government policy into account. What level of safety are the public, and by extension, governments, okay with? On average, there is a fatal accident every 100 million miles driven by a human driver.

Will driverless vehicles be held a higher standard?

The answer to that is going to play a part in determining whether driverless cars become more widespread and whether we eventually see Level 5 vehicles on the road.

We’ll see what the future brings.

The autonomous revolution is happening a little bit slower than expected, but it is happening. And it’s set to radically transform not only passenger vehicles, but long haul trucks, agriculture equipment, and so much more.