Blog Posts

Merging: The Latest Challenge for Self-Driving Cars

Dec 13, 2018

If you were to ask a group of motorists what their least favorite part about driving was, “merging” would probably rank high on their list. Apparently, self-driving cars think merging is tricky, as well. 

The challenge of merging

 When a human driver merges their vehicle onto the highway, they have to navigate the “person-to-person ballet” of driver communication, as Wired contributor Aarian Marshall put it. The human driver has to pick up on nonverbal cues like if an oncoming vehicle in the closest lane is slowing down to let them in, or speeding up to cut them off. In short, the human driver has to predict and understand what other human drivers are going to do.

As intelligent as self-driving car technology is, however, it’s still learning to accurately predict and respond to human behavior on the road. Sterling Anderson, co-founder of the autonomous vehicle startup Aurora, articulates what it’s like to merge from an autonomous vehicle’s perspective. “It has to identify where it is, what the intentions of the drivers around it are, and then perceive and track them through time and space for the opportunity to slot in.” And if the technology’s calculation is just a bit off, it can sabotage the merge (and possibly cause a collision).

Additional difficulties

And merges aren’t the only maneuver that self-driving cars find challenging. Passing through an intersection is even trickier since there are even more variables. Not only does the vehicle need to know its location in relation to other obstacles and vehicles, but it also needs to identify and respond to cyclists and pedestrians that might be crossing the road.

So, in short, self-driving technology still has a ways to go when it comes to merging and other driving maneuvers. However, considering just how far this technology has come recently, it’s likely that self-driving vehicles will master the art of the merge sometime in the near future.