Completed

A number of intelligent transportation systems (ITS) products and approaches have used “machine learning” for decades. Building on this foundation, recent artificial intelligence (AI) products and increased availability of AI have increased use cases for AI in transportation.

This project was conducted to introduce ENTERPRISE members to AI and facilitate understanding about how AI is being (and may be) applied to transportation, as well as the potential uses, benefits, and challenges of AI. Specifically, this project:

• Documents example use cases of AI in transportation operations. 
• Provides a definition and high-level context of AI compared to other solutions;
• Documents and synthesizes available national materials;
• Summarizes state-level AI policies to document themes and additional considerations;
• Identifies considerations for transportation agency practitioners interested in using AI in operations; and
• Documents example use cases of AI in transportation operations. 

Many vendors offer roadway sensors either embedded in pavements, non-intrusively within the road, or attached to infrastructure. These sensors gather data and/or communicate with vehicles to aid roadway operations. This effort seeks to understand existing and innovative types of commercially available intrusive and non-intrusive roadway sensors with an analysis of the potential applications, relevance, and drawbacks of each type. This effort considers sensors to be innovative if they are not widely used, either by being new to the transportation sector or not widely used by transportation agencies. Sensors of interest are examined in greater detail, such as identifying the ability of pavement to hold up structurally, operations and maintenance considerations, and placement of the sensors.

This Phase 2 ENTERPRISE Pooled Fund Study project builds on prior work conducted by selecting and advancing a managed set of sensor options from the Phase 1 effort. Specifically, this effort identifies possible test bed approaches that one or more agencies may test in a future test bed environment, with recommendations for common attributes of deployment (e.g., placement, security, communications) such that the individual deployments all contribute to a universal virtual test bed. Additionally, this effort develops a high-level systems engineering approach that defines the needs addressed, operational concepts, and preliminary requirements, with a goal of having a scientific-based description of the use of the roadway sensors identified in Phase 1 that can be presented internally by member agencies within their organization.

Wrong-way driving (WWD) is a growing concern because the resulting crashes tend to be severe and often result in fatalities and serious injuries. Advancements in cooperative automated technologies will enable significant possibilities to provide WWD warnings through in-vehicle navigation systems and smartphone-based mobile applications. A national communication standard for incident data including WWD events would enable data from multiple sources (e.g., 911 calls, on-road detection field equipment, self-reporting mobile applications) to become available through a data feed for in-vehicle mechanisms to access and provide alerts to errant drivers and nearby motorists.

This research completed a synthesis documenting several commercially available WWD in-vehicle systems and mobile applications, promoted the concept of a national communication standard and data feed for WWD events, and gathered input from transportation agencies regarding the readiness of agency-generated WWD event data to be contributed to a data feed. The project concluded that there is agency and industry interest in a national WWD communication standard and data feed. However, the WWD event data currently available from sources such as 911 calls, on-road detection equipment, and traffic management center observations is likely not yet ready to be pushed to such a feed. Future efforts to advance this concept can build upon national initiatives that are exploring the development of a data and communication standard for disruptive incidents, which could include WWD events.

Transportation agencies use many traditional data collection methods (e.g., manual counts, pneumatic tubes, in-road sensors, radar sensors) as well as emerging data collection methods (e.g., unmanned aircraft systems, probe data, video image detection and processing) for traffic data collection. While traditional traffic data collection methods have provided trusted data for many years, there may be advantages to using emerging traffic data collection methods to supplement or replace existing methods. For example, emerging traffic data collection methods may reduce resources needed to operate and maintain detection devices in the field, provide more complete coverage and faster access to data, or offer advanced analytics to increase an agency’s ability to glean useful insights from the data. Because tradeoffs likely exist when assessing various traffic data collection methods, it is important to compare emerging data collection options to traditional methods and learn from agency experiences. This ENTERPRISE project gathered traffic data collection methods and details through a literature search and survey of transportation agencies. After information-gathering was complete, a comparison of emerging traffic data collection alternatives to traditional methods was conducted and documented.