There are many compelling reasons to adopt connected and autonomous vehicle technology in our transportation systems. Safety is undoubtedly at the top of the list: Highway fatalities account for 93 percent of all U.S. transportation-related deaths and, as the Eno Center for Transportation points out, driver error is the main factor in over 90 percent of vehicle crashes. Vehicle automation technology, with its promise to greatly reduce crashes -- and even eliminate them completely -- would seem to be the ultimate safety solution.

To be sure, there are plenty of other potential benefits, among them reducing congestion, increasing mobility for the disabled and improving land use by, for instance, converting existing urban parking spaces to other uses. Self-driving technology, says a recent National League of Cities report, "could allow cities to redevelop at least 50 percent of their current street parking permanently."

That's the kind of thing officials in congested Somerville, Mass., had in mind when the city recently entered into a partnership with Audi, the German car company, to not only create self-driving cars that park themselves but also to build the infrastructure to accommodate the technology, as my colleague Tod Newcombe writes. According to Audi, 60 percent more self-driving cars can be squeezed into a parking garage because they need less room to self-park. The garage itself no longer would need to be located downtown, and existing parking structures and curbside spots could be freed up for other uses.

Yet, as is typical with new technology, there are significant issues to be addressed regarding deployment of self-driving cars, and by and large it will be governments that will be taking that on. The RAND Corp. has issued a pair of reports addressing these issues that are of particular value to state and local government officials. One is expressly directed at policymakers and the second is focused on how best to realize social benefits from the technology. Among some of the seriously tough problems that must be resolved before drivers can be fully removed from the scene: making sense of what's happening in the world around the car, operating under widely varying environmental conditions, assuring the reliability of the technology, finding the money for investments in infrastructure communication equipment, and dealing with the issue of cybersecurity.

  Another interesting issue also has been raised about decision-making on the part of robotic vehicles. In emergency situations, it turns out that we humans are often very intuitive in selecting avoidance measures. How might driverless cars be programmed to replicate or even improve on that? Who will make the car's safety programming decisions -- the carmaker, a standards organization or some other entity? And might there be different safety-level options available to a passenger in the vehicle?

Clearly a whole new dimension will be added to the responsibilities of government officials as they begin making regulatory, policy and administrative decisions to accommodate driverless cars. However, implementing vehicle automation isn't an all-or-none proposition. To assist in the transition, the National Highway Traffic Safety Administration has defined five levels of vehicle automation. These progress from no automation at all (Level 0) through "function-specific automation" (Level 1), "combined function automation" (Level 2), "limited self-driving automation" (Level 3) to the ultimate: "full self-driving automation" (Level 4).

Adaptive cruise control (Level 1), which automatically adjusts a vehicle's speed to maintain a safe following distance, is already showing up as optional equipment on cars. Tesla recently took things up a level when Model S owners received a software upgrade to their cars' autopilot mode (Level 2) that combines auto steer, auto lane change, auto park and side-collision avoidance. All of the major car makers are introducing variations of these technologies, and the trend will only continue to accelerate. (Google is taking a different tack and is focused solely on Level 4 technology, with the current Google car being the best-known example of a Level 3 vehicle.)

But while visions of a driverless future are enticing these companies and engineers, they can't be permitted to be in the driver's seat when it comes to this transformation to a new transportation system. It's been said that technology has no democratic imperative. Well, it can also be said to have no design imperative. For confirmation, one need only look at the car-centric development pattern of our 20th-century communities -- the result of letting horseless-carriage technology and its advocates run the show.

What people want is reduced congestion, increased mobility and a transportation system that makes communities more livable and prosperous. Civic and government leaders need to partner with the private sector, but they must responsibly guide and manage the integration of this promising technology to create a highly functional multi-modal transportation system.