The Road to Safe, Secure Driverless Cars


The development of autonomous vehicles promises a future of safe and efficient roads, unimpeded by distracted, impaired, aggressive, or deliberately speeding drivers. But to achieve this, the companies involved in developing driverless cars will have to navigate significant obstacles.

The transition from personally controlled to automated vehicles can be likened to the shift that occurred over the past 20 years from brick-and-mortar retail to e-commerce. For traditional storeowners, security depended on door locks, alarm systems, cameras, and access to cash registers. For online retailers, security has to do with networks and software.

Similarly, the safety focus in driverless vehicles will be largely about securing the networks and software that drive the cars. Today’s cars have approximately 100 million lines of code in them. Autonomous cars will have many times more. The companies that manufacture driverless cars will have to actively manage all of the security aspects of the vehicles’ software.

Today’s carmakers have, over time, developed efficient procedures for recalling and fixing vehicles with parts identified as faulty or unsafe. Similarly, with autonomous vehicles, manufacturers will need to devise methods of identifying and fixing problems discovered in software. In many cases, repairs can be done remotely, in the same way that mobile phone and computer makers can send patches over networks. But however fixes are made, management of software supply chains will need to be as efficient as the management of the supply chains for physical parts.

Beyond being efficient, software providers for driverless cars will surely face requirements to certify that the code they deliver is free of security vulnerabilities that, if exploited, could enable a hacker to seize control of the vehicle. A faulty spark plug is one thing. Suddenly having your steering, acceleration and braking hijacked is quite another.

Today’s driver feels in control. Indeed, the feeling of power that comes from being behind the wheel of a fast, stylish machine is one of the main selling points of cars today. With so many variables present in any driving environment, this feeling of control is, to a large extent, an illusion. Still, motorists’ willingness to hand over that control to software will depend largely on carmakers’ ability to gain their trust.

This is a difficult but not impossible challenge. Consumers have demonstrated they will accept change. Airline passengers today don’t seem to worry about automatic pilots guiding airplanes through the sky and even landing them when visibility is poor. Once upon a time, consumers would insist on personally talking to a salesperson and examining a product hands-on before buying it. Today, we are likely to purchase almost anything online.

We don’t know yet who the major players in the development of the driverless car will be. Candidates include the traditional carmakers, ride-sharing companies, rental car agencies, tech giants, tech startups, and—most likely—combinations of these players organized as strategic alliances. One early example is the recent announcement of General Motors and Lyft of plans for joint development of a network of on-demand self-driving cars.

The rise of autonomous vehicles presents important opportunities for growth for many companies. The winners will be those that can demonstrate to consumers that by ceding control of their vehicles, they will get the absolutely safest, most secure ride possible. The integrity of the software that controls the vehicle will be the key to success.

Lou Shipley is a Lecturer at the Martin Trust Center for MIT Entrepreneurship at the MIT Sloan School of Management. Most recently, he was CEO of Black Duck Software (acquired by Synopsys). Follow @loushipley

Trending on Xconomy