Self-driving cars have been getting a lot of buzz lately. Residents of Florida, Nevada, and California have already seen Google’s experimental cars cruising their highways, with sensor packages on their roofs and no one inside.
If the technology works in the “real world,” driverless cars are just around the corner, right? Well, maybe. Just because the test cars are licensed to operate in three states doesn’t mean they’ll be for sale soon. Some important issues are still unresolved but the final effect may be merely to shift the driver from operator to supervisor – a typical transition in most automated systems.
There are two reasons that driverless car technology is finally likely to happen: its apparent benefits and the amount of interest within the auto industry. Consider such likely advantages as:
- Safety – 90 percent of fatal traffic accidents have some human contribution. Computers are typically faster than humans at most tasks, so computer control could provide faster and more consistent responses to road conditions. And computers don’t get tired behind the wheel.
- Efficiency – 80% – 90% of road space is empty at any one time. Computer control could maintain safe spacing while reducing inter-car distances, yielding less traffic congestion and reduced pressure to build new roads
- Economy – Many modern cars already offer cruise controls that optimize fuel economy. A driverless car is essentially a continuous cruise control
- Convenience – Cars with automatic parking already exist. A driverless car could go one better,
- though, by dropping its passengers off and then parking somewhere else. The car sharing business model would also get a boost; simply call from a smartphone and have a car dispatched to you, (an effect that the taxicab industry might not welcome!)
- Lifestyle – Driverless cars could offer mobility – and richer lives – to aging or disabled people who might not otherwise have access to unscheduled transportation
- Productivity – Imagine being able to text, talk, web surf, or email without endangering yourself or others
Business interest in driverless technology is active and growing. Some big automotive players are involved, including:
- Google, by far the dominant force in technology development
- Tesla discussions with Google
- Nissan, which has committed to fielding a driverless car in less than a decade
- Mercedes-Benz, which is work with its own driverless vehicle design
Driverless technologies include a range of complementary or redundant sensors that feed data to a control computer.
The Google control system consists of a multi-beam laser range finder, radar sensors, a camera to detect traffic lights, inertial measurement, and a GPS. Sensor information is compared to stored maps that locate the car in space, augment the scene with any unusual objects, and then deliver control commands. It’s a complex system designed to operate in a complex environment.
So when can we expect to see these road warriors for sale? That depends. Estimates range from three to twelve years. That doesn’t seem too bad, but some non-technological issues have been raised that could push those dates out.
The biggest concern is liability. In other words, who’s responsible for an accident? Who pays for the consequences of a software or sensor failure? Who gets sued – the driver (a term that becomes fuzzy in a driverless car) or the manufacturer of a failed component? Although these legal issues aren’t hard to define, no one can yet say what the resolutions will look like.
At least there is guidance for one issue, however – the role of the human in this system. The concept of an automated car seems little different from the concept of an automated aircraft or an automated plant.
These systems involve human supervisory control, where people monitor the automation and only intervene when it fails. There’s a rich body of research behind supervisory control, which could provide a technical foundation for identifying failure modes, human response requirements, training and licensing standards for driverless cars. That foundation could, in turn, provide a framework for the legal environment in which these cars must operate. The car remains a tool and not a substitute for the driver and drivers are still responsible for the handling of their vehicle.
Driverless cars are another opportunity to examine the institutional structures that we use to foster the benefits of change and to protect us from its hazards. The solutions found for implementing this technology could help characterize our approach to large-scale innovation in general. And those solutions may show, yet again, that complex systems still require humans to ensure their success.