High-tech proponents envision a future in which car accidents have been largely eliminated by self-driving cars. But lots of nuts-and-bolts testing is required to determine whether and how that could happen.
Have you ever wondered, for example, why self-driving cars are most often tested in places like Phoenix and Southern California? Part of it may be the regulatory climate, but the most important reason is that those locations limit the scope of the experiment. How? They’re prone to sunny, dry weather.
A new generation of autonomous vehicles is being tested now in Boston, which exposes them to new variables like snow, rain and cold. A deep freeze can reduce the potential mileage of the battery in an electronic vehicle by 30 percent simply by keeping the electrons from being warm enough to operate efficiently.
Weather can also interfere with the vehicles’ sensors. “Snow not only alters the vehicle’s traction but also changes how the vehicle’s cameras and sensors perceive the street,” was the conclusion of a study by the Boston Consulting Group and the World Economic Forum.
Seagulls have also been a problem. They’re not bothered by the quiet electrical vehicles, and they can stop the vehicles simply by standing in the street. (So far, the sensors meant to stop the car from hitting obstacles like pedestrians have some difficulty distinguishing the size and type of the obstacle.) The engineers have now programmed the cars to startle the seagulls by creeping toward them.
When people imagine the barriers to a successful autonomous vehicle fleet, most picture the difficulty of getting a machine to react effectively to all the variables in driving — especially human variables. It is relatively straightforward to program the operating software with driving rules and traffic laws. But how do engineers expect computers to respond to human drivers who break the law or make mistakes in their driving?
According to Bloomberg Businessweek, those challenges are well on their way to being resolved. How to get the sensors to work in heavy weather? That’s still being studied.
The problem is that three of the four main sensors in an autonomous vehicle can be affected by weather. GPS can be slow and spotty in bad weather. Traditional cameras are ineffective in fog and snow. Lidar technology, which issues a laser beam to bounce off nearby obstacles, gets bounces from raindrops and snowflakes. That leaves radar, which cannot distinguish much between objects. With the sensor array basically flooded with misinformation, the vehicle can hardly establish where it is.
These challenges can be overcome, although engineers expect the process to take time. For now, a heavy fog or an inch of snow can blind an otherwise autonomous vehicle.
Whatever the solution to these and other testing problems, autonomous vehicle manufacturers should always put safety first. When they fail to do so, law firms like ours are here to hold them accountable for the injuries they cause.