ATSSA releases case study on AI uses in roadway safety

ATSSA releases case study on AI uses in roadway safety

Research from the U.S. and U.K. explores applications to help save lives

What began as a panel discussion at the Circle of Innovation at this year’s ATSSA Convention & Traffic Expo evolved into research that produced a new tool to help the industry understand ways to benefit from artificial intelligence (AI).

“Driving Transportation Safety Forward with AI: Case studies on the application of artificial intelligence in transportation,” was co-authored by ATSSA Director of Innovation & Technical Services Eric Perry and ATSSA Master Instructor Tim Luttrell and is now available as a free download.

It details research underway utilizing AI including studies conducted by the Iowa Department of Transportation (IowaDOT) on AI for traffic incident management, work by the Hawaii Department of Transportation (HDOT) on a safety analytics platform, pavement marking condition assessments that take advantage of its capabilities and research on planning data applications.

The ATSSA case study also includes additional examples in the early stage of development for advancing roadway safety such as:

• pothole detection from dash camera images at Loyola Marymount University;
• applying AI with camera data to automated pavement marking applications in Pennsylvania;
• advancing situational awareness through automated video incident detection and other machine learning tech in California;
• evaluating a fuzzy logic ramp metering algorithm in the Seattle area;
• AI for incident detection in Nevada and Florida; and
• construction applications from a tech firm in the United Kingdom.

The case study notes that transportation agencies and the private sector are looking to improve efficiency in how they are utilizing data and recognize AI has the potential to “think, learn and process data in much more efficient ways than traditional methods.”

“Practitioners see AI as an integral component in the paradigm shift from evaluating historical safety data to proactively saving lives through such technology applications,” the authors state in the introduction.

They note that application of AI falls into five macro-level categories: text AI, visual AI, interactive AI, analytical AI and functional AI. And they point out that in its advanced stages, AI can be utilized to analyze safety data, in predictive ways and in decision-support tools. Such information can then be used to impact decisions on issues from “assessing pavement markings to developing enhanced asset management programs to more efficient maintenance programs.”

Research already underway has resulted in considering the following:

• What if AI could predict when and where traffic incidents are most likely to occur?
• Can AI be used to evaluate maintenance needs in place of time-consuming site reviews?
• Does AI have both in-vehicle and roadside or roadway technology capabilities?
• Is AI capable of running innovative pavement marking machines to maintain existing markings?
• What types of data can AI use to help with infrastructure management and maintenance?
• Can AI also communicate with in-vehicle technologies for a broader positive impact?
• Could AI provide inputs to our team that help us save lives instead of reviewing lagging statistics after lives are lost?

The case study ends with a summary of benefits already being seen from AI research.