Why We're Still Facing A Long Road To Autonomous Vehicles

Every year, $25 trillion worth of stuff gets made, moved, and sold around the world. I've devoted my career to studying this global product economy, where so many fascinating stories are hiding in plain sight. One of those stories is the uncertain future of self-driving vehicles. In a recent interview with Re/Code, Elon Musk said that he's confident we'll see mass production of self-driving vehicles within 15-20 years. A pretty short timespan, considering the amount of work that needs to get done. Living in Mountain View, home to Google and its self-driving initiative, I'm used to sharing the road with autonomous cars. But analysts are still arguing over how long it'll be before the average consumer sits behind the wheel.

I see autonomous vehicles as the first step to a new kind of society: totally redesigned cities that forego traditional frustrations of traffic and congestion. But to get there, we need to get better at marrying the eagerness of software with the reality of hardware. Building products is hard. There are a lot of nitty-gritty details we need to understand before we can utilize the full impact of self-driving vehicles.

Tech Minded
It's no surprise that the leading pioneers of self-driving cars are software-oriented. Google and Tesla, the two frontrunners for the development and proliferation of autonomous vehicles, have always been about technological moonshots. A number of other companies have also joined in the race to create a car that doubles as a chauffeur -- such as Ford shifting from a car to a mobility company -- and so we're presented with a number of different types of technology.

Each company differs when it comes to their intended path. Google is approaching self-driving technology head on -- their strategy will likely mimic their approach to Android technology: licensing software to other hardware manufacturers. Google uses LiDAR technology, a complicated system of lasers and sensors that, as of this writing, is the most accurate form of autonomous tech. But it's expensive, and would require mass production to be affordable to consumers. Tesla, on the other hand, is taking a different approach by perfecting autopilot tech before branching out to totally autonomous vehicles. Recent news of the first passenger death in a Tesla may have deterred some, but advocates of autonomous tech emphasize that accident rates in a self-driving world will likely be far lower than they are today. Tesla's use a mix of sensors and cameras to recognize other cars and lane markings -- but the software is not as advanced as LiDAR, and cannot yet fully "understand" every kind of driving environment. As Tesla continues to iterate, we will likely see a move towards full autonomy.

Autonomous cars don't have to go the way of the HD DVD vs Blu-Ray battle. Two or more strategies might be able to coexist, offering varying levels of autonomy. There are a number of ways the software in autonomous cars could find a place in mainstream society, but we won't know the best option until one company commits to building the first mass-produced autonomous car.

But here's the thing: making cars is especially difficult, because people are going to be putting their lives at risk every time they get inside. It's not enough to develop the technology -- companies need to develop technology that can be mass produced. Whereas software can be reprogrammed easily, consumer products need to be built to perfection every time. I know, I know, this is obvious information. But it's not always so obvious when it comes to what will take a product from stuff of dreams to stuff in our homes. In the case of self-driving cars, we're going to have to figure out software issues, as well as the data that drives it (pun intended).

Data Dilemmas
One of the most daunting speed bumps facing self-driving cars is the issue of data. How is it being collected and secured -- and who owns it? The success of self-driving cars will rely on lightning-fast communication between vehicles that necessitates the rapid storage and dissemination of data. While companies like Huawei in China are developing 5G networks that will allow data to travel quickly, car companies are already arguing over what happens to that data.

Our driving patterns tell more about us than our social networks do -- we unconsciously reveal deeply personal information just from driving to and from our decided upon locations every day. So if that data were to get hacked, or sold to advertisers -- what then?

The real key to self-driving cars hitting the mainstream is solving the myriad of complex problems that data collection and communication present. Whoever can solve those problems stands to influence an entirely new era of vehicles, especially since automakers will once again have to align on one process in order to ensure compatibility across models.

How to Build
Once the issues of data and software have been finalized, the task of building the car will present itself. And that's going to make up half the battle.

Barring the initial cost of hardware, the way self-driving cars are manufactured is going to determine their future success. Software is easy to iterate on. Products -- especially cars -- not so much. Manufacturing is going to take a long time to get right, and it's going to be costly. For car companies to build reliable and lasting self-driving cars, they're going to have to marry the exciting world of software with the years of expertise behind car manufacturing.

What I'm talking about is what many are calling the Third Industrial Revolution: a huge shift in the way we build products influenced by software and data. A deviation from the traditional world of manufacturing into a new, fast-paced universe, tapped into all sorts of sensors and programs that allow companies to build products faster and better. The way we choose to build self-driving cars will define how this industrial revolution plays out into the future -- specifically into the Fourth Industrial Revolution (also called Industrie 4.0), which will be driven by highly localized manufacturing. Companies like Tesla are already redesigning factories to allow for this innovative production.

That's the exciting thing about self-driving cars, especially from a future-of-products perspective: if companies can get this right, then we're looking at a huge shift in the way we build everything.

Giving Up Ownership, Not Freedom
A big reason we're all so excited about self-driving cars is the prospect of less congested streets. But contrary to popular belief, vehicle autonomy won't actually do much to ease up our overcrowded freeways.

Unless we give up ownership of our cars.

Elon Musk imagines a future where someone can hail their car from hundreds of miles away. Great for parking issues, not so great for traffic -- what do we imagine freeways will look like if there are cars on the road with zero drivers? Just as carpooling logically reduces jams, adding driverless vehicles to the road will simply clog them. The only solution is to create a world in which we no longer own cars. A world where urban centers are completely reshaped by a more universal, connected transportation system.

Uber has invested research into self-driving vehicles, and many executives from the company and others like Google have expressed their belief that car ownership will disappear as autonomous -- likely electric -- vehicles take over. Uber imagines a future where people hail self-driving cars for their day-to-day needs, never having to own a vehicle.

As much as I love the thrill of driving fast down isolated roads, I'd gladly give up ownership of my vehicle if it meant getting to work quicker, reducing the risk of accidents, and a cleaner environment. We'll see fewer gas stations and parking lots, making room in cities for housing and public spaces.

As for 15-20 years time? If companies can solve the issues of data communication and technology (not to mention government regulation), I can absolutely see autonomous vehicles garner wide adoption within the next two decades. And, who knows? The successful leap in automation could spark a large-scale product revolution.

Just don't underestimate the art of making stuff.