Each year, Texas highways capture several tons of road debris, said Yan Huang, a UNT professor of computer science and engineering.
“At high speeds, even small pieces of tire can cause sudden vehicle swirls and secondary collisions and miles of traffic jams,” she said.
Now she and her colleague Heng Fan are working with TxDOT to tackle the problem: They’re trying to develop an artificial intelligence system to quickly detect road debris and alert crews.
Right now, Huang said, TxDOT doesn’t know anything about debris on roads unless it’s reported by someone.
“You do it after the incident has happened and even hang there for a while, causing chain reactions and secondary crashes,” she explained.
Huang hopes to combine several data sources, including the WAZE app, where people can report objects on the road, to help detect hazards in real time.
“We have seen that roughly 72% of debris reports come from WAZE, and these reports are typically received about 16 minutes earlier than traditional methods,” Huang said in a November press release. “By combining WAZE with other crowdsourced data, we hope to detect debris even faster and improve response times.”
They will also use TxDOT closed-circuit cameras.
“We are integrating multiple data sources, including crowdsourced data and the connected vehicles, dashboard camera data, which is becoming more and more available,” she said.
Fellow researcher Heng Fan said the system does not store personal information such as human faces or vehicle license plate numbers.
“So that means the AI system will always be safe to detect just the debris,” Fan said.
Huang said the Texas A&M Transportation Institute also wanted to reach out as a partner for this project.
They hope to launch a prototype powered by one mission next year.
“On busy highways, minutes really save lives. So this is the goal of our project,” Huang said.
