Auto executives and technology experts have been touting the impending arrival of self-driving cars for almost a decade. So then, where are they?
To start, as automated vehicles grow closer to fruition, their Jetsons-esque appeal is becoming less fantasized, and more hard-hitting. In other words, reality is setting in. Many alarming events have already occurred, such as in March 2018 when one of Uber’s self-driving cars in Tempe, Arizona, struck and killed a woman. Additionally, since 2016, four Tesla drivers have been killed while using the company’s semi-autonomous “autopilot” feature.
Whether autonomous vehicles are actually going to make cities better, or make them vastly worse, is also a burgeoning question; and one that starts with the concerns of the individual citizen. Many worry that self-driving cars will bring along with them a whole new set of problems and questions. Take, for instance…
Computers can malfunction. Many people are nervously reluctant to hand over all the power to a computer, that, if malfunctioned, can put the driver in a more dangerous situation than if the driver were operating the vehicle themselves.
It’s not yet clear how self-driving cars would maneuver through unexpected hazards like roadblocks, or how they will be able to discern unique local driving laws. For instance, different states have different laws regarding turning right on red. Will the computers could have difficulty identifying these different local and state road rules?
Self-driving cars aren’t the do-nothing luxuries they’re made out to be. While the computer takes over once the vehicle is operational, the passengers would still be required to maintain some knowledge about how to operate it safely. How will this education be disseminated?
Some are concerned that the cost of implementing the new technology will be out of reach for most Americans. Currently, the engineering, power and computer requirements, software, and sensors add up to more thank $100K. Will automated cars become yet another luxury that further perpetuates classism?
Many are not grasping the reality that in order for self-driving cars to attain their promise of better safety, things may likely first become less safe. A majority of people will need to “opt in” to the service in order for the most savings in terms of cost, time, and lives to be fully realized. If the service is not adopted widely, accidents can and will still happen.
Another major obstacle is that, because of the high reliance on technology for automation, the very security of these vehicles could likely be of high interest to hackers… a terrifyingly dangerous potentiality.
Due to the vast amount of information that would need to be stored on the software in order for a computer to operate a vehicle, many are concerned about their personal data being collected.
Another not-so-talked-about con about self-driving cars is that they’re expected to eliminate many jobs in the transportation sector, particularly when it comes to freight transportation and taxi drivers. This could have a negative impact on not just the unemployment rate, but the economy as well.
Legal concerns abound due to lack of legal precedents. A self-driving car doesn’t completely eliminate the likelihood of car accidents, so when accidents do occur, how would these cases be handled? Who would hold responsibility – the driver, the car manufacturer, the software developer?
There are also ethical concerns. In a case where a self-driving car needs to avoid a car crash by swerving, but to do so would mean potentially steering toward the sidewalk where a person happens to be, how do engineers program it to decide?
Machine-learning systems all suffer from the same fundamental, unsolved problem: inductive reasoning. Machines are built to analyze millions of data points and find patterns; in other words, they are built for deductive reasoning. To teach a machine how to recognize squirrels, for example, it is shown a million squirrel photos. After some time, it learns to distinguish between “squirrel” and “not squirrel.” However, it cannot reason inductively – the ability to predict whether a squirrel on a sidewalk will stay there or dart into traffic. Or, whereas a human can easily recognize a stop sign with a tree branch hanging across it, the branch could cause a car’s computer vision system not to see the stop sign at all.
Automated vehicles are not able to operate at the highest levels of safety in all weather situations. Common complications like rain, fog and shadows can cripple even the most advanced autonomy systems. In fact, heavy rain alone can seriously damage the laser sensor mounted on the car’s roof. How large a role then would a passenger have in the event the technology unexpectedly fails?
If other technology fails, such as traffic signals, the current model employs human traffic cops to step in and direct traffic. However, automated cars cannot interpret human signals, thus leaving open the potential of danger for all involved.
The reliance on technology could mean that, over time, drivers lose their skill to operate regular cars. In the event of a technology glitch or recall, out-of-practice drivers could be rendered helpless to get around.
The success of self-driving cars currently relies on accurate GPS mapping systems. But anyone who drives can attest to having, on occasion, been advised by their GPS to turn down a one-way street going in the wrong direction, or told that they were driving on a non-existent street. Without a human driver to course-correct, where does this leave the safety of a self-driving car’s passengers?
With Google currently at the helm of development for the cars, other auto manufacturers might sell fewer cars as people hold out for the Google version.
It’s likely that self-driving cars, being touted as “new and improved,” would be electric. This threatens the gasoline industry quite a bit, which is expected to suffer.
In summary, tech gurus and automotive executives have spent a lot of time focusing on the positive aspects of self-driving cars, such as the potential for reducing traffic and emissions, the freeing up of parking spaces for sidewalks and cafes, and the potential for achieving near-perfect safety records for drivers and pedestrians thanks to advanced cameras and quick-as-lightning reaction times. But more should be made clear about the many challenges ahead for autonomous vehicles becoming a standard means of transportation. For the self-driving car industry, getting around these challenges is certainly proving to be an, erm, obstacle course. So it may be awhile longer before self-driving cars hit the road running – and we’re going to say that’s a good thing.
