Imagine a vehicle that backs itself out of the driveway, waits for you at the curb, knows where you’re going and the best roads to get there. You get into the car, but instead of driving, you’re the passenger. Instead of navigating rush hour traffic, you spend your commute time eating breakfast, working, or watching a movie. When you reach your destination, the car lets you out at the curb and finds its own parking spot or continues on to serve its next passenger.
This futuristic scenario is fueled by the vision for fully autonomous vehicles, and behind the scenes there is a fierce race among automakers and technology companies to be the first to develop this capability to define the future of mobility. Some $80 billion had been poured into the development of autonomous vehicles through June 2017 by traditional automakers and a range of other interested parties, including the big tech companies, and competition among those involved is ferocious, according to The Brookings Institution. While it remains to be seen who will win, all of the players face massive challenges in five core areas: artificial intelligence (AI), data management, the legal front, public safety and the ability to scale autonomous mobility technology.
Before digging deeper on those challenges, it’s worth reviewing where we stand today.
Test cars from the likes of Google’s Waymo, as well as Lyft and Uber, have traditional steering wheels and look and feel like a typical car. But the nearly autonomous vehicles that begin to appear on our roads will offer passengers an experience more akin to riding on a commuter train. Instead of having two people in the front and two in the back, autonomous vehicles will likely accommodate four people facing each other over a work table. They might also offer spots where a person can sleep comfortably.
In fact, we’re moving to a time when owning a car may not be the best option for many people — and that puts tremendous stress on a business model that’s existed for the automakers for more than 100 years. Especially in Europe, travelers may need a car for only a small piece of a multimodal journey that includes bus, rail and even a service such as Lyft or Uber.
The automakers look to pull ahead
While companies such as Tesla and Waymo have grabbed an early lead in driverless cars, the traditional automakers see the writing on the wall and have been investing heavily, lining up partners and pouring resources into development.
Every major automaker in the world wants to be an early mover, according to Brookings. And if they are not vying for early mover status, they strive to avoid competitive disadvantage. The automakers have been joined by leading automotive suppliers such as Bosch, Continental and Delphi Automotive.
But road tests and simulations create petabytes of data and result in massive data challenges. To cope, many automakers have leaned on computer science companies to solve these data issues.
The tech companies are also joining the race. Along with Waymo, Apple, Microsoft, Alibaba, Baidu and Tencent are increasingly in the mix, and other tech players such as Intel, NVIDIA and Qualcomm are investing big money to make the microprocessors needed for autonomous cars. Fleet operators are also investing, including rideshare and logistics companies that are likely to be the early adopters. These companies have a ready supply of vehicle miles to contribute data for machine learning applications that will change the future of mobility and transportation.
So, while there has been a lot of speculation about how the autonomous car development race will play out, the participants are still jockeying for position, and we can expect to see more deals and partnerships. Already, Toyota has invested $500 million into ridesharing company Uber to pilot an autonomous vehicle. And Honda and General Motors announced a partnership to build autonomous cars together. Other partnerships include Argo AI and Ford and Waymo and Chrysler.
The tech companies may claim they can build a better autonomous car, but to really produce cars at production-level scale, they’ll have to partner with the automakers and the industry’s vast network of suppliers and systems integrators.
The bottom line: It’s unlikely that any one company will go it alone. The automakers may have production experience, but they’ll also need AI and next-generation computer science and data management skills to build better autonomous cars more quickly.
Five challenges to master
Given the task at hand, here are five major challenges the autonomous driving industry must master to achieve success:
1. Artificial intelligence
The automakers must look at AI as a new discipline and lean on systems integrators and the broader computer industry for expertise. They will need AI platforms to manage the flow of data at petabyte scale and AI experts to make sense of it all. This will require recruiting talent across the globe who can work in flexible and agile teams.
They will also have to merge AI with internet of things (IoT) technologies and leverage that marriage going forward. Already, edge computing devices close to the testing action are being integrated into IoT environments to give AI the performance data required for continuous improvement in the near future. To obtain road approval for future innovation, AI technology and driver knowledge must be managed safely and traceably.
2. Data management and privacy
Test fleets can generate petabytes of data per day around the globe. The automakers need to get the right datasets in front of the right research and development (R&D) teams for the right situations being tested.
For example, an R&D team may be testing how vehicles react when pedestrians suddenly cross the road from the right side. The team needs only the videos for that specific situation, not the massive number of frames from the vehicles in all the other scenarios the cars are put through.
Then there’s the consumer data challenge. The new cars will track data on everything from the number of turns the car makes to where the car’s passengers dine, where they shop and where they work and travel by car on vacation.
Automakers need to figure out how to use this data to help consumers while grappling with the obvious privacy policy issues.
3. Political and legal implications
Different countries will have to work out whether trade associations and independent standards groups will regulate the rules for autonomous cars, or if it will be done by government agencies. An entire new field of case law and automobile insurance will also develop when inevitable accidents and other product liability disputes arise as autonomous cars make their way into the market.
In the United States, state governments have slowly been granting permits to test driverless vehicles. Waymo, for example, has been testing driverless vehicles in Phoenix, Arizona, and California has given Waymo the first permit to test driverless vehicles on public roads without a test driver present. The company has begun testing in Los Angeles, Los Altos Hills, Palo Alto, Sunnyvale and Mountain View.
The Chinese government granted permission to Daimler to test driverless cars on public roads in Beijing. Daimler also has licenses in the United States and Germany, and has already announced a pilot project to start offering an autonomous shuttle service in San José, California.
4. Public support and safety
When cell phones first emerged, signal coverage was poor. While that frustrated consumers, there’s much more at stake with autonomous cars. As accidents involving autonomous vehicles occur, the industry will have to spend a great deal of effort to get the public to take the long view on the potential for this burgeoning technology: Over a decade or more, autonomous cars are expected to significantly reduce road fatalities and deliver more convenience to private citizens and business travelers alike.
In many ways, this is an article of faith. Right now there are no studies, and the historical data does not exist to prove that autonomous cars will be safer in the long run. So the industry has to remain transparent and report to the public frequently about its progress with next-gen data and AI technology, and its testing of autonomous vehicles. As automakers prove they can build safe, reliable autonomous cars, the public should grow to trust and accept the vehicles.
5. Scale
Traditional automakers are less concerned with which company gets on the road first with near autonomous vehicles, as opposed to which company can mass-produce autonomous vehicles at scale. Making that happen will require the automakers to move out of startup mode and into the next phase, where they identify the building blocks that can scale on an industrialised level. Data management will be critical. Building autonomous cars at scale will require standardised data-gathering methods and an open ecosystem for exchanging data and algorithms for verification and validation, so the automakers and their development network and partners can deliver continuous improvements as the technology for advances in autonomous driving development.
Teaming for success
The major automakers may be slightly behind, but they are confident they will have a profound story to tell and will prevail long-term. They certainly can’t be ignored. The car industry has many interdependent suppliers and, as autonomous cars evolve, these relationships will become even more important. Traditional automakers and their suppliers can build cars at scale, and that will be critical if they hope to deliver autonomous cars at price points people can afford.
But the automotive industry must consider the big picture: How can all these complex automotive and computing skills be mastered on a production level so the industry can deliver better quality at less cost for consumers?
The tech companies, on the other hand, have to ask themselves whether they’ll be able to keep up or if it would make more sense to forge strategic alliances with the automakers. The next-generation computer science companies and service providers play a crucial role in the autonomous car race because they have the necessary computing skills in their DNA to push automakers and their development partners to the next level of autonomous driving.
In the tech business, however, the trend for many years has been for a dominant player to emerge in each market — the Microsoft desktop, the Cisco router, the Google search engine. However, it’s unlikely that the autonomous car market will be dominated by one or two companies. There might be dominant regional players in the United States, Europe or Asia, but there is too much at stake and too many viable companies working on the issue to presume that only a few will survive.
Instead, it seems likely that there will be a meeting of the minds and skills as the autonomous car market evolves rapidly. The next few years promise to be very exciting. Expect to be amazed.
Learn more about DXC's Automotive expertise, and contact us to continue the conversation.
About the Author
Matthias Bauhammer is offering director and global lead for DXC Robotic Drive. He is responsible for the thematic development and strategic projects of DXC's artificial intelligence and analytics programs in the automotive industry.