The auto industry is working hard to ensure that the future of fully autonomous cars is not that far down the line, but a new online simulator neatly outlines the challenges carmakers must face to get there.
A fully autonomous car, or Level 5 autonomy, means the car can drive itself without human intervention. This would make those images we’ve all seen online, of people sleeping at the wheel or reading books, a concrete and safe reality. We’ve still got a very long way to go until then.
This week, the Washington Post released an online simulator that highlights the challenges current testing AVs face. For the record, the industry has only reached Level 2 or 3 autonomy and great strides are needed to get to 5.
AVs are fitted with a plenty of sensors, LiDAR and cameras, which collect data and feed it into the car’s computer. Based on this data, the car has to make decisions that for human drivers come naturally and very easily through experience. The car is also expected to predict behavior on the road from a multitude of agents, including other drivers, pedestrians, bikers and even animals. And this is where today’s AVs fail.
As the simulator shows, AVs have issues identifying animals smaller than a moose. They also can’t distinguish accurately at all times between a pedestrian and a cyclist, and even have trouble seeing a car broken down in the middle of a lane.
Ultimately, that’s the biggest problem AVs are facing right now: they can’t “see” and “hear” properly, in addition to their inability to predict behavior. A human driver sees 4 times farther in daylight than current technology. The technology is rendered useless by extreme weather as well, like fog, heavy rain or snow, or nighttime.
“What confounds the vehicles also complicates efforts to develop uniform standards among companies and establish regulations for safety, insurance rates, liability and traffic laws,” the Post notes. Federal regulation on AVs is needed for safety reasons, Adam Scow, a senior advocate for Consumer Watchdog, tells the publication.
This week, the Washington Post released an online simulator that highlights the challenges current testing AVs face. For the record, the industry has only reached Level 2 or 3 autonomy and great strides are needed to get to 5.
AVs are fitted with a plenty of sensors, LiDAR and cameras, which collect data and feed it into the car’s computer. Based on this data, the car has to make decisions that for human drivers come naturally and very easily through experience. The car is also expected to predict behavior on the road from a multitude of agents, including other drivers, pedestrians, bikers and even animals. And this is where today’s AVs fail.
As the simulator shows, AVs have issues identifying animals smaller than a moose. They also can’t distinguish accurately at all times between a pedestrian and a cyclist, and even have trouble seeing a car broken down in the middle of a lane.
Ultimately, that’s the biggest problem AVs are facing right now: they can’t “see” and “hear” properly, in addition to their inability to predict behavior. A human driver sees 4 times farther in daylight than current technology. The technology is rendered useless by extreme weather as well, like fog, heavy rain or snow, or nighttime.
“What confounds the vehicles also complicates efforts to develop uniform standards among companies and establish regulations for safety, insurance rates, liability and traffic laws,” the Post notes. Federal regulation on AVs is needed for safety reasons, Adam Scow, a senior advocate for Consumer Watchdog, tells the publication.