We need to plan for this reality and ensure driverless cars deliver concrete safety gains under these conditions.
In our prior Post on autonomous vehicles, we suggested that driverless cars may not be safer than humans in certain situations because they may not observe certain human driving behaviors or draw the correct inference based on those behaviors. For those who have embraced a driverless future, this is a temporary safety risk that will dissipate as autonomous vehicles increasingly dominate the roads. So how long will human drivers remain on the roads? We don’t think they are going anywhere anytime soon.
The vision for driverless vehicles is grounded in the sharing economy: Cars are expensive and underutilized; therefore, it would make more sense to share them so that they are always in use and ownership costs are spread across many users. In the future, this will be even easier to accomplish with driverless vehicles that are constantly “on call.” People will subscribe to car services or own time shares in driverless vehicles in order to meet their transportation needs without the headaches of monthly car payments, insurance, and operations and maintenance costs.
This vision may be attractive to many, but it requires one essential ingredient: Population density. Many people may be willing to ditch driving if an available car is never more than a minute or two away. But if your ride is 5 minutes away or more, the case for going driverless weakens substantially. If it is more than 15 minutes away, very few people would be willing to abandon having their own car. This means that in rural America, the driverless car revolution is unlikely to take hold, at least not as part of the sharing economy.
In the suburbs, people may be unwilling to rely on driving as a service for a variety of other reasons. It may be less efficient than driving your own car when running multiple errands. If you have children with car seats, it would be incredibly inconvenient and potentially unsafe to be constantly removing and installing the car seats in different cars. Similarly, how viable is driving as a service if you need a special roof rack (e.g., bike, ski, or surf) or an accommodation for your pet?
Of course, in rural America as well as the suburbs and cities, people could own their driverless cars rather than share them. They may be willing to pay for the ability to go driverless whenever they want to, such as during their daily commutes. But that does not mean that every person who owns a driverless car would want to go driverless in every scenario.
Indeed, there are many people who enjoy driving (someone has to buy the cars that still come with manual transmissions), and there are already driving enthusiast groups forming to ensure we have steering wheels in the future, such as the Human Driving Association. And there are many others who would trust an autonomous vehicle in certain circumstances (e.g., cruising on the highway during a commute with lots of other driverless cars) but not others (e.g., driving down a suburban residential street where lots of kids are constantly playing, riding bikes, walking dogs, etc.). This means that there will likely be cars driven by driving enthusiasts for the foreseeable future, and there will likely be hybrid autonomous vehicles that can switch from being driven to driverless based on driver/occupant preference.
It seems quite likely, then, that we need to plan for a future where driverless cars will be sharing the roads with drivers. It is therefore critical that autonomous vehicles be able to operate safely when they encounter humans behind the wheel. We’ll be on the lookout for independent testing that addresses this issue in a variety of contexts. Stay tuned.
Thanks for Driving2Save.