Opinion: A winding road to self-driving cars
By Paul Witman, Jim Prior and Scott Mackelprang
Many of us will reach an age or condition when we can no longer drive. It would be wonderful if we can get technology to do the driving for us so we can still have the same freedom.
But such vehicles aren’t here yet. And current versions still have quirks, like the Waymo vehicles that wrongly routed dozens of trips into a dead-end cul-de-sac in San Francisco, disrupting traffic for residents.
The road from cars requiring drivers to autonomous vehicles can best be understood using the Society of Automotive Engineers classification system, with levels 0, 1 and 2 Advanced Driver Assistance Systems (ADAS) requiring a driver, and levels 3, 4 and 5 Automated Driving Systems (ADS) ultimately leading to a vehicle without a human driver.
Many ADAS features are available in cars today, from early solutions like adaptive cruise control, forward collision warning, automatic emergency braking, lane-departure warning, lane-keeping assistance and blind-spot warning to more advanced lane-centering technology, lane-changing assistance and self-steering systems. Drivers may already be developing dependencies on these capabilities and becoming complacent because of them.
However, there are currently no ADS vehicles available for consumer purchase, though some systems are being tested for use in taxis and other such commercial operations.
There are many potential social and economic benefits from existing ADAS features as well as the future ADS state of fully autonomous vehicles: increased safety for drivers and passengers along with other motorists and pedestrians; increased mobility via expanded transportation options; reduction in car crashes; less need for parking lots and spaces; increases in automated ride sharing reducing environmental impact; and reduced traffic congestion.
Of course, there are challenges as well: liability and insurance changes with no driver to hold accountable; a need for new regulations; a lack of industry standards; and technology limitations that make some conditions harder to manage, like fog, rain and poor road markings.
There are also anticipated risks of malicious hacking and the potential for privacy issues, as the vehicles coordinate their movements and share their observations about road conditions among themselves. And all of these concerns have to be appropriately weighed against the economic and social value of entrepreneurship so that we don’t over-regulate.
There are also user-related risks. Vehicle drivers are not as highly trained or regulated as pilots who use automation devices on aircraft, nor do they have a copilot on board. Drivers risk becoming inattentive, as our minds don’t like to watch for the rare danger when comfortable with an illusion of general safety. Drivers may also be more inclined to risk “driving” while intoxicated, reasoning that the car will do most of the driving.
Drivers have many times bypassed alert-driver detection systems, subverting those controls, although vehicle owners explicitly agree to a set of responsibilities and risks in their user agreements.
These new risks can endanger other drivers, pedestrians, first responders, cyclists and road workers. Even though an ADAS car may be statistically safer than the average human driver, there still have been cases of those vehicles behaving erratically in unexpected ways that sometimes require the humans to take control. These occurrences may result in damage as well as injury or death to the ADAS driver and to those around them.
It is incumbent upon lawmakers and regulatory bodies to appropriately protect the public as a whole. Legal liabilities should be clear and appropriate for all the many scenarios. Regulators should carefully consider the full range of risk trade-offs and require or implement appropriate compensating controls to offset that. Those controls might include limiting the conditions in which and areas where these systems can be used or other constraints that manage the risk to the public. Automakers need to be crystal clear about, and actively enforce, the need for driver attention.
Only then can we safely get from “I drive” to “it drives.”
• Paul Witman is an information technology management professor and Jim Prior is an adjunct instructor at California Lutheran University. Scott Mackelprang is a consultant with expertise in information security, privacy and risk.