The mass production of the Ford Model T was a truly revolutionary development to land-based transportation in the early 20th century. The “horseless carriage” has obviously been a worldwide success.
For a variety of reasons social and otherwise, the allure of driving no longer always maintains its shine. Whether for short trips around town or long trips across the country, there are times when a land-based form of autopilot would be preferred.
While many have dreamed of driving automation for decades, a few companies are now pouring serious resources into autonomous car development and gaining remarkable traction.
That these cars are not only in development, but on public roads in live test mode should be a sign to both citizen and legislator alike that it is now time to seriously and thoroughly ask questions about how driverless vehicles should be deployed among us.
Much of the media coverage to date on self-driving technology has not yet narrowed its focus to public safety and the in-depth implications of these vehicles operating in our many forms of public traffic. One high-profile article was even titled, “How Google Got States to Legalize Driverless Cars.”
The following questions are offered in the interest of stirring public discussion and debate on these technology developments.
Will we allow people to have their driverless cars drop passengers off and go park themselves? Can your driverless car be your own valet? Even if we stay in these vehicles until they are parked, what about parking enforcement? After having paid for our spot, we may inadvertently overstay our time in a metered parking space. What if driverless cars figure out how to evade parking enforcement officials? Should using that app on our driverless car be illegal?
How open will driverless car development be? That is, will there be an app store where developers can write apps for the car that integrate sensors, customize radio programming or movie playing based on time of day, lightness outside, location, etc.? What if there's an app to evade parking enforcement officials to have the car re-enter traffic and find another parking space (and notify the passengers via app of the new location)? Will there be apps for changing driving behavior near speed traps?
Would driverless cars recognize law enforcement vehicles? Will driverless cars know they should pull over for law enforcement when the lights are on or turned on? (Note: This, of course, can happen at 25 mph or less, too.) Will it know the difference between the different color lights (police vs. fire)? What about when the colors mean different things in different states? Will driverless cars change their behavior (such as speed) when they notice law enforcement vehicles nearby, moving or not, with their lights off? Would a driverless car know how to get out of the way of emergency vehicles?
Would a driverless vehicle accurately observe and know how to avoid other vehicles that are moving too fast, or even too slow, in its current lane or another lane nearby that may not be within range of its sensors?
How would car-sharing services make use of driverless vehicles? Would they be allowed to have driverless cars move their empty selves on to the next user? Would a car-sharing service dispatch a shared driverless vehicle in between parts of another user's commute? How would these kinds of service options affect the locally regulated taxi industry and its new higher end competitors like Uber?
For any of these times when someone thinks it makes sense to have an empty driverless car driving on the road (parking, vehicle sharing, etc.), what happens when the ratio of empty cars in the public roadways becomes noteworthy to other drivers? For every algorithm that finds ways to reduce traffic, there are plenty of unintended consequences that could jam up the roads even worse.
What about when a driverless car app developer thinks it's useful to have cars make use of residential streets to get around traffic? Will driverless cars know how to read signs that do not allow traffic in certain directions at certain hours to prohibit rush hour avoidance? Does a driverless vehicle know how to slow down for speed bumps?
Much of driving consists of “keeping your eyes on the road.” We see many things on and near public roadways, and respond accordingly.
Would a driverless car “see” a train? What about the crossing guards before and after a train passes? If the crossing guards are stuck blocking traffic, would a driverless car see traffic management officials and know how to move accordingly? What about those rural crossings with no crossing guards?
Speaking of rural, does a driverless car know what a farm tractor looks like? They use the public roadways, too. Will driverless cars know how to navigate dirt roads, especially the curvy ones on hills? What about roads that have not been striped yet?
What about small children who sometimes play near roadways? Would a driverless car slam on the brakes if a child jumps out in front of it? Can it also recognize people on bicycles? If it were to, God forbid, hit a bicyclist or pedestrian, would it know that, and would it actually stop? If it did not and the car ends up being the perpetrator of a hit-and-run, who is responsible? Will vehicle manufacturers or their software developers have legal immunity?
How would driverless cars handle adverse weather? How much rain would reach a level of interference with a driverless vehicle's sensors that would render it unsafe to operate? What about snow, ice, sleet, etc.? Does a driverless vehicle know how to handle a hydroplaning situation? How exactly would one of these vehicles handle a self-determination that it must pull over and stop to wait for severe weather to pass? Can a driverless vehicle tell when the road ahead of it is flooded? Does it know how to “Turn around; don't drown”? What about when dirt roads turn into mud? What about if there's a drop-off ahead? Who wants to make the first off-road driverless vehicle?
Driverless vehicle development may not end up being all that open to developers. Nonetheless, what if the security of the operating system or the software operating it is compromised? As wireless Internet access becomes ubiquitous and driverless cars become part of the Internet of Things, what happens when a driverless car's destination gets hacked? Don't count on anyone in the car to notice as one main benefit of a driverless car is to be able to ignore the ride altogether. If passengers end up in a dangerous location, who is responsible and liable?
As vehicles and our technologies become increasingly interconnected, the amount of data generated, shared, and stored also increases. What kind of privacy rules should be in place regarding data on where people and their vehicles go? What kind of access to data should owners have for driverless vehicles that could end up stolen? If one thinks biometrics would prevent that, there are others who would find ways to compromise that in one way or another.
The FBI has also expressed concern over possible criminal uses of driverless vehicles.
How will insurance companies handle increased liabilities from driverless vehicles? There's not exactly a track record of probabilities for the actuaries to establish trends. State legislators regulate insurance, also, so that adds a second area of jurisdiction.
Many of the questions here related to basic safety are not much more complicated than the tests every licensed driver must pass before independently taking to the public roadways. Another source for questions legislators should ask is our driving tests, written and in practice.
Can driverless vehicles practice defensive driving on the roads that we expect of ourselves? We must remember that things which are common sense to humans are not common to technology and the computer programs that operate them.
Some states also require safety and emissions inspections of vehicles. How do those change with the additional sensors and systems in place that enable vehicles to operate themselves? How much will those inspections cost? (This is asked on behalf of both those charging and those paying since the other rates are also set.)
Driverless vehicles are vastly more complicated in the name of making driving simpler for people. This raises the stakes for these systems to operate perfectly much higher than when a human driver is operating a vehicle and is responsible at all times for all motion of the vehicle. Should states require separate licensing for operating driverless vehicles? Should users be required to regularly inspect the sensors, especially the visually-based ones, for obstructions? How often? Is the software on driverless vehicles required to notify users of obstructions and interference to its sensors? How is the software certain there is or is not an obstruction to the sensors it uses? What kind of maintenance do these additional sensors and systems require?
Some of these sensors are radar-based. Is there any radar detection involved that may run afoul of state laws banning radar detectors? Could any of these systems interfere with law enforcement use of radar technology?
When NASA and its commercial partners launch a rocket, there is an extensive set of Go or No Go procedures they go through before launch. Driverless vehicles will also be making a lot of determinations and decisions based on information it receives. How much access to that information and those determinations and decisions will be accessible to those in the vehicle, especially those who want it? What about those remotely outside the vehicle? (Google is a company that collected Wi-Fi information as it took pictures up and down neighborhood streets.) How open should driverless car operation be? Should operating information be required to be human-readable?
Some of the manufacturers may already be asking these questions or others like them. It is still important for legislators to ask them and enshrine into law common expectations, even if virtually everyone already agrees to them now anyway.
I stand by those recommendations, and think the questions here demonstrate a need to more thoroughly examine the future of self-driving vehicles.
As one who both likes to drive both short and long distances, I should note that despite these many questions, I would actually like a self-driving car. There are many times where the road not needing my full attention would be convenient. My intent is to make sure we do this well. Driverless cars may very well be the next generation of cruise control, as one lawmaker put it. Driving automation should be about giving drivers and vehicle operators more control over their transportation experience, not less.
Driverless vehicles, especially those without user controls, could ultimately go the way of the Segway, with limited or specialized uses. Nonetheless, even amusement parks have their risks, too. If operated only in closed or mostly closed environments, lawmakers would still do well to ask similar questions about any vehicle interacting with or near people.
8/1/14 Update: This post was mentioned in a
1 comment:
best post
Post a Comment