Repost Wednesday

Repost Wednesday: How can carmakers adapt to driverless vehicles?

Repost Wednesday is a short series of guest-posts that will be featured on our blog periodically, highlighting some of the wonderful posts MIT Press authors have written about their books on other platforms. We hope you enjoy this series.

Carmakers can adapt to driverless vehicles by partnering with software companies—but may end up following their lead

When human drivers let intelligent software take the wheel, driverless cars will offer billions of people all over the world a safer and more convenient mode of transport. In an ideal future, our roads will glisten with swarms of tightly packed driverless cars. Moving in schools like fish, these cars will demonstrate extraordinary anti-collision abilities, effortlessly navigating through streets full of pedestrians and falling gracefully into fuel-efficient formations on long, empty stretches of motorway. Some cars will carry a passenger or two. Others will be empty, on their way to drop off a pizza or to pick up a child from school.

The exact configurations vary, but most driverless cars today use several digital cameras, a radar sensor, and a laser-radar (lidar) device to “see” where they’re going. These cars pair a global positioning system (GPS) device with another location device called an inertial measurement unit (IMU) that compensates for GPS inaccuracies. An onboard computer takes the information streaming in from the sensors and GPS, folds that data onto a high-definition map that contains stored information on crossroads and traffic lights, and processes it into a 3D model called an “occupancy grid.” Software knits all this data together and provides driverless cars with artificial vision that is increasingly accurate and responsive.

The epicenter of automotive innovation has moved to Silicon Valley. For the first time, traditional car companies face competition from software giants, namely Google. Driverless cars are disrupting an industry that for decades has operated inside protective walls, sheltered from external competition by high barriers to entry and protected by exclusive relationships between big car makers and preferred suppliers. So far, car companies have favoured an evolutionary approach, selling vehicles with gradually increasingly autonomous software modules that help human drivers with specific activities such as parking or driving on the motorway. In contrast, Google’s strategy has been to create fully autonomous cars that have no steering wheel or brakes, with no human driver needed or even desired.

As Google’s cars improve and consumer interest in driverless vehicles grows, the fate of car companies is unclear. Creating software capable of artificial intelligence—especially artificial perception—requires an army of skilled personnel and a depth of intellectual capital. While adept at creating complex mechanical systems, car companies lack the staff, culture and operational experience for AI research. To survive, carmakers will have to re-envision their products not as vehicles built for humans, but as intelligent and autonomous robots, a shift that will demand major changes in their workforces and development processes.

Car companies can scramble to develop in-house software expertise, or they can form a go-to-market partnership with a software company that will provide an operating system (OS), while the car company builds the vehicle’s body. We predict that a series of corporate “marriages” will arise, in which car makers will sell cars that are basic building block platforms to downstream software companies, who will transform them into autonomous vehicles. The risk, of course, is that as driverless cars become commercially viable and today’s tech-enamored teenagers mature into tomorrow’s car-buyers, a vehicle’s hardware—chassis, engine, and interior—will become a secondary feature to its software.

The situation in the car industry is reminiscent of the battle between software and hardware companies in the personal computer industry. We might describe this as the “Microsoft paradigm” versus the “Apple paradigm.” In the Microsoft paradigm, a software-savvy company (most likely Google or perhaps Uber) would control the market for driverless vehicles. Car companies would make cost-effective hardware while their more visible software partner outfits the “naked” vehicles with an intelligent OS. The software company would also serve as the production hub, installing and testing the car’s OS and managing the hardware vendors that provide the sensors. Essentially, the car companies would be demoted to mere equipment manufacturers, nearly as invisible and interchangeable as the bevy of companies who manufacture hardware to support the Windows OS.

If, as research predicts, the future market for personal car ownership shrinks, the Microsoft paradigm will prevail. If most self-driving cars are fleets of driverless taxis owned by intermediaries, a vehicle’s most distinguishing and valuable feature will be its software. Vehicle “hardware” will be marketed to fleet owners as a nearly generic commodity, yielding only a slim profit margin to car companies, while the sale of the more valuable product, the car’s software, yields a substantially fatter profit margin, reaped by the software company.

But not all consumers of the future will want to ride around in a generic taxi pod. In the Apple paradigm, car makers will sell their product directly to consumers who want an expensive, brand-name model, perhaps for a specific purpose. Consumers who purchase these models will be a desirable demographic: high-net-worth individuals. While a partnership with a software company will still be required, the car-maker will dominate the deal and sales will be about the quality of the whole car. If the Apple paradigm prevails, car makers will remain in control of an autonomous vehicle’s development and sales process, including the profits.

Google retains some major advantages as the undisputed industry leader in digital maps and deep-learning software. It is also eager to reduce its dependence on selling ads, still its primary source of revenue. Ironically, the car industry’s emphasis on building great hardware might end up being what keeps them in business. Driverless cars will still require a safe, fuel efficient body that meets exacting regulations. Unlike car companies, tech firms such as Google have no experience in assuming responsibility for the physical safety of the general public on a large scale. As car and tech companies begin to play their high-stakes game of poker, it remains to be seen who will have the winning hand

This post was written by Hod Lipson and Melba Kurman, authors of Driverless: Intelligent Cars and the Road Ahead, published by The MIT Press.

This post originally appeared online for Prospect magazine on May 19th, 2017.