Artificial intelligence (AI) may seem like it is taking over the world, but some researchers fear AI in transportation may be falling behind.
To support the future AI in transportation, a team of researchers led by Abhijit Sarkar of the Virginia Tech Transportation Institute (VTTI) wrote a report published by the National Academies of Sciences, Engineering, and Medicine. The report explores the benefits and implementations of AI in departments of transportation (DOT) at the state and local levels and provides a set of guidelines and research agendas.
“AI and machine learning are exciting fields. There is a lot of excitement around how these can make the transportation field more efficient and safer,” said Sid Mohan, associate program manager for implementation and innovation at the Transportation Research Board, a division of the National Academies that worked closely on the report. “Everything has to start somewhere, and this project could set the stage for something big in the integration of AI into DOTs.”
The report identifies 11 research problem statements that can serve as guidelines for departments to implement AI into their systems. Each statement comes with a list of action items, a proposed budget, and a timeline to facilitate a seamless implementation. The topics range from improving safety and traffic congestion to enhancing organizational efficiency and customer service.
“There is a paradigm shift in how AI is influencing ground transportation and transportation research,” said Sarkar, the team lead for computer vision and machine learning at VTTI. “This is the perfect time to embrace the power of AI and use it for the benefit of our society, improve road safety, and enhance transportation efficiency, reducing accidents, and saving lives on a global scale.”
The findings are the result of a comprehensive literature review from over 65,000 articles, complimented by expert interviews and information gained through workshops. In total, 29 individuals from eight state transportation departments were interviewed and a combined 56 individuals from state and local departments participated in two workshops.
“Our goal was to develop a research agenda related to the use of AI,” said Matt Camden, a senior research associate at VTTI. “However, we are not the ones out in the field. Because of that, it was critical for us to collect data and talk to the people who are using these technologies daily.”
Results showed departments are looking to integrate AI in areas of traffic management, safety, mobility, asset management, infrastructure, multimodal transportation, and pavement. However, there is a general lack of understanding about the technologies and a lack of an experienced workforce, according to the report.
The report aims to start filling this knowledge gap and provide a road map to integrate AI in transportation that will improve nation’s transportation infrastructure. Researchers believe that if AI was integrated into existing transportation systems, it could reduce the risk of crashes and enhance traffic management by predicting traffic flow and adjusting traffic signals to reduce congestion, among other improvements. Such improvements in safety are central to VTTI’s mission.
“Most transportation departments struggle to effectively integrate AI into their network and workflow,” Sarkar said. “Through this project we hope to provide them with a starting point.”
Virginia Tech
