| Abstract |
Transportation and public works agencies manage extensive networks of roads and supporting assets using information collected through fleet operations, inspections, and routine field activities. While these activities generate large volumes of imagery and observational data, much of it remains underutilized due to the effort required for manual review and the limitations of periodic, project-based surveys. As a result, actionable awareness is often incomplete, reactive, and disconnected from day-to-day operational decision-making.
This presentation examines how generative artificial intelligence (GenAI) is being applied to transform fleet-collected imagery into usable information that supports transportation and public works management. Building on applied experience from governmental environments, including deployments using AI-powered smart camera technology and platforms such as CITYROVER, the session explores how AI agents can be assigned defined review tasks to automatically analyze images captured during routine operations and convert them into structured outputs suitable for human review. These outputs include draft asset inventories, summarized inspection observations, identification of common condition issues, and preliminary maintenance prioritization.
Rather than relying on infrequent network-wide surveys or standalone digital twin initiatives, this approach emphasizes continuous data collection through everyday fleet activity combined with automated analysis. GenAI enables multiple review objectives to be addressed simultaneously using the same datasets, allowing agencies to extract greater value from information they already collect. The presentation will illustrate how this model supports transportation assets such as pavement, sidewalks, curbs, signage, drainage features, and vegetation, while remaining adaptable to local operational priorities.
The session will focus on practical implementation considerations observed in early deployments, including ideal data collection conditions, designing effective prompts and analysis pipelines, and how AI-generated outputs can be validated and integrated into existing asset management, inspection, and work planning processes. Additional topics such as cross-department information sharing will be discussed, with attention to how outputs can support coordination between operations, engineering, and asset management teams.
In this presentation, we will discuss key topics related how road and transportation authorities are moving beyond episodic data collection toward more continuous awareness, the limitations encountered in practice, and how similar approaches may be applies across different departments. We will be joined by the City of (TBD) to demonstrate new capabilities leveraging generative artificial intelligence to extract insights from digital twin imagery. We will demonstrate how different types of analysis can support different business unit needs.
(A few cities can co-present)
|
