Lesson

Take a piecemeal approach to the integration of road weather prediction models, starting with the easily available sources of data and slowly building up capabilities of a modular system.

Integrated Model for Road Condition Prediction (IMRCP) prototype provides a framework for the integration of road condition monitoring and forecast data to support decisions by travelers, transportation operators, and maintenance providers.


12/31/2017


Background (Show)

Lesson Learned

The IMRCP system is a unique research effort that brings together different transportation operations disciplines together to achieve a lofty goal of providing integrated predictions of road conditions to operators. Bringing together weather, road temperature, and traffic models together in real-time requires a complex data integration architecture coupled with big data management and effective visualization. The evaluation approach and findings provided insight into not only the successes in meeting the lofty goal but also the continuing efforts that are needed to bring IMRCP to a stable operational use. Below are key lessons learned that focus on future implementation for sites that might be considering the use of systems like IMRCP.

The quality of IMRCP results is dependent on foundational systems at the TMC.
    The IMRCP tool is based on a foundation of traffic data available at a TMC. These include detector data and event data typically available as part of an ATMS suite. In addition, IMRCP requires a representation of the network that is easily available. If there is an existing micro/meso simulation, verified detector data and robust tracking of events at the TMC, the set-up of IMRCP is greatly simplified. However, in the absence of verified foundational data, the start-up hurdles of setting up IMRCP can seem daunting. However, it is important to note that agencies can (and likely take) a piecemeal approach to IMRCP integration starting with the easily available sources of data and slowly building up capabilities of this modular system.

Expand focus beyond the workday.
    A key lesson learned from the evaluation was the need for the IMRCP system to be set up to account for the "unusual" since those conditions are where operators greatly value the improved situational awareness. Traditional model development focuses typically on calibrating for the frequent, predictable workday conditions. Counter-intuitively, this is when operators feel that they can manage the event using their existing tools. Weekends, special events, and nights are often times when non-recurring congestion (and associated impacts) occur that are not as well monitored by current TMC systems.

Invest in Calibration and Testing.
    While calibration and testing were a core element of the project, the evaluation revealed the need for more robust approaches to calibration and testing before operational deployment. In essence, as the first generation of the tool, the evaluation period served more as a "shakedown" period with many errors and limitations coming to light as operators, evaluators, and system deployment teams started looking more at the output results. While the test plans focused on component level testing, additional end-to-end testing could have revealed some of the issues earlier.

Reporting Exceptions Rather than the Norm.
    The IMRCP tool was also a first in figuring out how to assemble the various data and model components in the system for use by a TMC operator. As the evaluation progressed, it became clear that "less is more" in terms of the information provided to the operator. For systems like IMRCP, reporting the "exception from the norm" is particularly important. For example, the notification feature added by the system development team was well received by the TMC operators. Additional effort is needed to enhance the notification of exceptions including multiple exception handling, prioritization and user interface displays. It is important to note that, as a research tool, the IMRCP project was only scoped to develop a functional user interface rather than spend resources on a polished user interface. (See the following lesson for additional detail.)

Encourage Direct Integration of IMRCP Outputs into ATMS.
    Fundamentally, the IMRCP is a decision support tool that ingests data and provides notifications of current and forecast conditions. While a separate user interface was developed for this project, the more realistic use case is to build a data interface directly into the ATMS. Data feeds from the IMRCP can directly feed into event notification systems at the TMC and operators can be made aware of a change in conditions directly on their screen as opposed to having another browser open with this information. Once the interface is built, additional capabilities become possible such as direct dissemination to travel information systems or to support messaging/paging of TMC/maintenance staff when alert thresholds are met.

Develop Scenarios of Use for Traffic Prediction.
    Lastly, the use of weather forecasts in decision-making is intuitive. In fact, people use weather forecasts in their daily lives. Road weather information such as pavement condition and pavement temperatures are also well-understood by the maintenance community and they are able to adequately adjust their strategy and tactics based on forecasts of such information. On the other hand, the role of traffic prediction in TMC operations is still emerging. Most TMCs rely only on current data for decision-making. To adequately integrate predictions into the TMC set up, additional work is needed to establish scenario managers that translate the traffic prediction to more meaningful actions for the TMC. For example, a variable speed limit scenario manager might take into account traffic predictions and alert the operator that a speed change might be needed in 15 minutes.


Lesson Comments

No comments posted to date

Comment on this Lesson

To comment on this lesson, fill in the information below and click on submit. An asterisk (*) indicates a required field. Your name and email address, if provided, will not be posted, but are to contact you, if needed to clarify your comments.



Source

Integrated Modeling for Road Condition Prediction

Author: Garrett, J. Kyle; Hani Mahmassani; Deepak Gopalakrishna; Bryan Krueger; Jiaqi Ma; Fang Zhou; Zihan Hong; Marija Ostojic; and Nayel Urena Serulle

Published By: ITS Joint Program Office Office of the Assistant Secretary for Research and Technology U.S. Department of Transportation

Source Date: 12/31/2017

Other Reference Number: FHWA-JPO-18-631

URL: https://collaboration.fhwa.dot.gov/dot/fhwa/RWMX/Documents/IMRCP/IMRCP_Final_Report%2012-20-17-FOR%20508.pdf

Lesson Contacts

Lesson Analyst:

Kathy Thompson


Rating

Average User Rating

0 ( ratings)

Rate this Lesson

(click stars to rate)


States

None defined

Countries

None defined

Systems Engineering

Show the V

Unit / Device Testing

Focus Areas

None defined

Goal Areas

Productivity

Keywords

None defined

Lesson ID: 2018-00824