KALAMAZOO, Mich. — Autonomous vehicles are expected to account for about 12% of car registrations by 2030. Western Michigan University students hope to be a part of the bettering technology to make that happen.
Their research focuses on how autonomous vehicles drive in inclement weather like snow. They're hoping through data collection. They can better computer systems to pick up on things like snow accumulation, lane lines, and object detection.
“Since I started driving here, the first problem I faced was during a snowstorm, my car, it was so difficult to drive,” said Anika Tabassum, a Western Michigan University computer engineering undergraduate student.
It has been a little over a year since Anika Tabassum learned how to drive since moving to Michigan from Bangladesh.
With snow on the road, that was an even newer concept for her.
“That’s when I started to think about what happened; even for a human like me, it was tough to control the tires and to take control of the car myself,” said Tabassum.
That made her question how autonomous vehicles react in the same conditions.
“Most autonomous vehicle systems rely on clear conditions to localize the vehicle better. They need to see their environment without any exclusion,” said Parth Kadiv, a Western Michigan University Mechanical Engineering PhD student.
The students are conducting their research through Western's Energy Efficient and Autonomous Vehicle lab.
They're now collecting thousands of miles of data to see how these systems can run in different weather events.
“Usually, when it starts getting snow or foggy or raining, those are typically the conditions where the sensor degradation starts, so the sensors do not perform as well,” said Kadiv.
“The differences in weather can make the algorithm completely different. The parameters change, so that is when we have to figure out how to deal with those parameters and those variables to take extra steps,” said Tabassum.
With lots of snow on the road, they say many of those systems currently fail.
The students hope their research can improve what's already out there.
“I think even if we could address even a little bit, like 5% of it. We need these systems to work when it is not safe outside because they are meant to keep us, you know, safe, especially in conditions we cannot control,” said Kadiv.
They aim to launch this technology onto an embedded chip and commercialize it.
The project, funded by the National Science Foundation Partnerships for Innovation, runs until 2025.