Researchers at North Carolina State University are working to make driverless vehicles safer, by running computer simulations.
Merging onto an interstate is old hat for experienced drivers. But how do you teach that skill to a computer-driven autonomous vehicle?
Assistant professor Ali Hajbabaie has been wrestling with that question for years. He says computer models that simulate real-world conditions are the answer.
"We decided to add different maneuvers like now, you want to go from one lane to the other one, just because you want to avoid something in front of you, or you want to reduce your travel time," Hajbabaie said.
Hajbabaie and his team, which includes students, tries breaking these problems down into small algorithms, which are easier for computer programmers to write into code.
Hajbabaie says these simulations are needed to make sure autonomous vehicles are safer and more efficient than those driven by people.
"So that not only it's absolutely safe, but also it gives the best travel time to the vehicle," he said.
The next step is testing the code in small model replicas of driverless vehicles. A real-world tryout is still years away.
