Countless automotive and software companies are burning the midnight oil to bring self-driving cars to the market. While some of them have projections for release in the early 2020s, some industry experts are saying that may not be the case. A generous timeframe puts autonomous car releases at around 10 years away, whereas some believe we have decades to go before we see the technology perfected.
Self-driving vehicles have many obstacles still to overcome, as shown by the fatal Arizona crash involving an Uber autonomous vehicle. These challenges will take time to resolve as the brightest minds come together to build promising autonomous vehicles. In order to move past the hurdles, researchers have some work to do before cars can be trusted without human intervention.
Ever since the aforementioned incident involving the death of a pedestrian at the wheel of Uber’s autonomous car, many drivers are reluctant to hand over control to a smart vehicle. A survey conducted by AAA found that 73 percent of Americans would be too afraid to ride in a completely autonomous vehicle. That number is up from 63 percent the previous year.
Vehicle companies have begun giving passengers autonomous car tests so they can experience it for themselves. According to Waymo’s Krafcik, after an initial skeptical period, most riders resorted to playing on their phones, even falling asleep as the car navigated.
Varying weather conditions give rise to a whole new set of issues. Autonomous cars rely on cameras and sensors to navigate over roadways. If snow blocks the pavement, the car sensors can no longer view lane lines. Additionally, heavy fog and rain have the potential to disrupt the view of cameras.
Light beams used to locate obstacles could mistake snowflakes as obstacles in heavy snowfall. Radar is capable of seeing through the weather, but it still wouldn’t be able to make out shapes necessary for computers to discern what it’s seeing.
Researchers are currently working on laser sensors that utilize an alternate light beam wavelength to see through snowflakes. At the same time, software is in development for vehicles to differentiate between real obstacles and weather-induced confusion.
Interestingly enough, human drivers pose a rather interesting challenge when it comes to autonomous vehicles. Whereas the cars will be programmed to follow a certain set of guidelines, humans are not. Meaning they are still capable of running red lights, double parking, walking in front of moving cars, blocking intersections, and the bane of every traffic-logged person, being cut off.
Humans are also capable of observing other drivers, making eye contact, and giving another driver a sign to go ahead of their vehicle. Autonomous vehicles don’t yet have this capability.
Roadways vary from place to place. Lane lines aren’t standardized, so some may exist in one location and in others they don’t. Human drivers can make this alteration in their driving depending on where they are and the visual cues presented to them. Intersections and curves are also high on the list of obstacles for autonomous vehicles. Depending on where the lines are placed to stop, an autonomous vehicle may not be able to see buildings or traffic at the cross.
Left turns seem relatively straight forward, but in certain situations, they are somewhat like a dance. Making a turn in front of oncoming traffic is something that requires proper timing and good decision making. Humans find making left turns where there are no green arrows, rather difficult, resulting in hundreds of accidents each year. Autonomous vehicles find the same task just as problematic.
While researchers and industry experts may be at odds about when we’ll see autonomous cars take over, we do know that we’ll see them at some point. Whether it’s sooner rather than later, that is still up for debate.