Self-Driving Cars

Cars that can drive themselves are inching closer to reality. Google driverless vehicles have passed the 500,000-mile mark on public highways without a driver, and have sustained no crashes (a better mark than many cars with drivers). With many of the technical challenges overcome, it's time to consider the associated policy issues. A recent study by the RAND Corporation attempts to tackle these issues.

The RAND group interviewed groups from all perspectives-regulators, automakers, communications groups, and others-and concluded that the overall benefits to society are likely greater than the potential problems created by autonomous vehicle technology.

It is intended that this study help to guide policymakers as they grapple with the practical issues created by this new technology. For example: should this be regulated at the state level, federal level, or both? It seems likely that some regulation will exist at both levels, but some states and the District of Columbia have already produced some laws regarding the use of driverless vehicles. Regulation must be carefully crafted to avoid stifling the technological benefits.

Those benefits are quite significant, according to the study. Assuming the technology works as intended, crash reduction is an obvious benefit. With efficient communication between vehicles and control systems, commuting time should decrease and congestion should be reduced, consequently saving energy and reducing pollution. Populations with reduced mobility (the blind and otherwise disabled, for example) will have new opportunities. Collective efficiency may be increased as people can do other tasks while commuting. Autonomous cars can drop off and park themselves, reducing parking requirements.

Of course, there are corresponding potential detriments. Correctly interpreting foreign objects in the roadway is not straightforward for autonomous vehicles. The efficiency and congestion assumptions may prove to be false if more people are encouraged to "drive" by the newfound ease of a commute. The interdependent nature of a smoothly flowing autonomous traffic system may produce massive chain-reaction accidents from a software or communications error. Security will be extremely important as hackers can cause significant, immediate damage.

Further technological challenges include building relatively economical fail-safe, redundant systems to handle things like sensor failure-for example, a sensor that is sending no data is easy to dismiss, but what about a sensor sending incorrect data? The best analogy is the control system used in manned spaceflight, but redundant systems of that nature aren't economically practical for commercial vehicles. Who will decide how safe is safe enough?

Considering all these factors, the Rand group makes four basic recommendations:
• Resist the urge to regulate before the technology is viable; let it evolve.
• Determine how to update distracted-driving laws to accommodate self-driving vehicles.
• Address the privacy issues raised by the new data streams from self-driving vehicles, and establish ownership.
• Compare performances of self-driving vehicles to average drivers when considering regulations and liability assessment, and take into account the long-term benefits when considering liability (in other words, don't let potentially catastrophic events scuttle the whole technology without a serious risk assessment).

Many years are likely to pass before you climb into your self-driving vehicle and take a relaxing commute to work while reading the paper or watching a movie...or, for you Type A personalities, doing work before getting to work. Even so, this study should help to pave the way for an event-free commute when the time comes.