Institute for Transportation

About the research

Furthermore, to account for the relatively limited data volume, the researchers established a transfer learning framework, which greatly eliminated the need for training a large number of hyperparameters. The transfer learning algorithm achieved a precision of 88% using daytime optical images and an impressive precision of 94% when using nighttime thermal images, despite the constraints imposed by using a limited dataset.

About the research

State agencies spend tens of millions of dollars on winter maintenance each year. For example, the State of Indiana spends upwards of $60M annually on salt, fuel, labor, equipment, and other maintenance costs, so it is imperative to make data-driven decisions. Advanced crowdsourced connected vehicle data have emerged in the past eight years that can leverage beyond weather data, vehicle dynamics data from engine output, and drivetrain and wheel sensors. Micro-slippage and roadway friction can be estimated at 75-ft segments of roadway aggregated at a 10-minute frequency without any additional instrumentation, from consumer vehicles off the production line. Traditionally, agencies have leveraged RWIS and other road weather sensors for tactical decision-making, but they are expensive to deploy and maintain and can only provide limited spatial coverage. Agency maintenance vehicles such as snowplows run on dedicated routes, and the timing of deployments, as well as the length and duration of routes, may not be representative of general traffic behavior. A crowdsourced solution provides more agile and broader network coverage because of the moving nature of vehicles.

This research includes a set of tasks to evaluate commercially available connected vehicle (CV) data to measure friction, wet-state, and ambient temperature over large road networks across multiple states. Static RWIS can be used for groundtruth if CV data is gathered in the proximity. If the new data source is found to be usable, a contingency plan on how these data can be integrated into the existing datasets, decision-making systems, and business processes will also be developed.

About the research

Road weather systems are used by state and local agencies to mitigate and manage the disruptive impact of weather events on roadways. Some of the fundamental aspects of road weather systems are the collection of weather-related data from environmental sensor stations and probe vehicles, the processing/distribution of data, and the determination of how/when/where to deploy road maintenance resources and/or to issue general traveler advisories and/or issue location specific warnings to drivers.

As momentum behind connected vehicle technology continues to build, practitioners are showing interest in determining how connected vehicle technology can be leveraged to support traffic management activities, including road weather systems. Specifically, the ability to communicate with connected vehicles opens up new opportunities for collecting data from many vehicles, and targeted dissemination of information to drivers. Thus, it will be important to ascertain the types of data that can be communicated in connected vehicle messages, as well as other intrinsic aspects of connected vehicle communications to understand how connected vehicles can enhance existing and open up opportunities for new road weather strategies.

Research will be undertaken as part of this project to review connected vehicle data standards, and to engage Aurora members to determine which road weather strategies are of greatest interest to practitioners. The project team will apply knowledge gained from members, as well as their background in engineering systems, to develop a Concept of Operations, which will provide a description of how connected vehicle communications and data may be employed to enhance the capabilities of road weather systems.

About the research

Adverse weather conditions present significant risks to motorists, making safe navigation challenging. AI advancements have facilitated the development of intelligent systems to address these concerns. This project introduces CARWIS (Conversational AI for Road Weather Information Systems), an AI-powered solution that provides real-time data on road conditions during severe weather. CARWIS gathers data from various sources, including weather forecasts and traffic cameras, and employs natural language processing to generate timely and accurate insights into road conditions. CARWIS can detect hazardous conditions, such as icy roads or low visibility due to fog or precipitation, and refine its predictions over time, enabling drivers to make informed decisions on their travel plans, potentially reducing accidents and enhancing safety. Additionally, CARWIS can assist transportation professionals in planning and responding to severe weather events by providing detailed information on road conditions. This enables officials to prioritize resources and make well-informed decisions regarding road closures and safety measures. This innovative solution harnesses the power of AI to improve road safety and reduce incidents during adverse weather conditions. As AI technology continues to progress, it is anticipated that more advanced systems will be developed to assist in navigating and managing severe weather events on the roadways.

About the research

The lack of a national standard for winter weather road condition indices has led to inconsistencies in assessing road conditions and providing accurate information to drivers across the United States. This research aims to develop a national standard for winter weather road condition indices that is consistent, accurate, and reliable, enhancing driver safety and winter weather response effectiveness. The project involves conducting a comprehensive literature review, data analysis, and case study analysis, as well as engaging key stakeholders from public and private organizations. The anticipated outcomes include a national standard framework, implementation guide, training materials, monitoring and evaluation toolkit, best practices repository, communication templates, and an interactive map/dashboard for road conditions. The implementation of a national standard for winter weather road condition indices is expected to improve driver safety, reduce traffic crashes and congestion, and optimize winter weather response strategies by transportation agencies.

About the research

Variable speed limits (VSLs) are useful in promoting highway safety. Along these lines, the Federal Highway Administration (FHWA) mentions, “the use of VSLs during inclement weather or other less than ideal conditions can improve safety by decreasing the risks associated with traveling at speeds that are higher than appropriate for the conditions.”

The goal of this proposal is to automatically recommend speeds for various weather conditions (rainfall, snow, ice, fog, etc.) at roadway segments that are good candidates for VSL. This means that the roadway segments should frequently experience adverse weather conditions (such as snow,

rain, fog, etc.), high traffic, or safety hazards. The crash rate at such road segments should generally be higher than average. The research team expects to gather road weather information system (RWIS), traffic, friction, incident, and potentially other data sets over one or more seasons that typically exhibit adverse weather. The team will then utilize the collected data and develop analysis methodology in establishing VSL algorithms that consider different terrain types, roadway geometries, and weather conditions (rainfall, snow, ice, fog, etc.). The team will explore the usage of machine learning (ML) algorithms and other approaches in establishing VSL. The speed limits will be set to satisfy the driver’s visibility and stopping sight distance requirements and also prevent lateral slippage at curved sections considering the loss of friction due to inclement weather conditions.

About the research

Deciding when to remove spring load restrictions (SLRs) on roadways is complicated given the variable time window during and after thawing when excess moisture remains in the base and subgrade layers, causing the overall roadway structure to remain weak. The main objective of this project was to develop an economical and easy-to-use protocol for timing SLR removal.

To develop the model, the research team utilized falling weight deflectometer (FWD) data from three test cells at the Minnesota Department of Transportation’s (MnDOT’s) MnROAD research facility. FWD data from nine other sites were used to validate the model, with three sites in North Dakota, three in New Hampshire, two in New York, and one in Maine. Numerous statistical analyses were performed on the FWD data sets, and model/protocol development considered factors such as base layer and subgrade type, effects of moisture, and depth to the groundwater table.

The researchers created a decision tree to help agencies implement the SLR removal guidelines developed in this study. To use the decision tree effectively, it is necessary to know information about the roadway structure, base layer(s), and subgrade soils and the approximate depth to the groundwater table. Using this methodology may help transportation agencies lift their SLRs more quickly than they have in the past.

About the research

Until recently, road weather information system (RWIS) stations have been overlooked and underutilized by many transportation agencies and maintenance personnel. It may stem from the fact that the data is point measurements with large spatial gaps in between, resulting in an incomplete view of road surface conditions (RSC) over the entire highway network. Also, one of the most important pieces about RSC, namely, snow coverage or bare pavement status, which is available from RWIS cameras or fleet cameras, is only accessible manually by maintenance personnel or road users. Therefore, there is a need to automate the recognition of snow coverage via images from RWIS or other data sources such as fleet dash cameras or traffic cameras, improve RSC inferences at unmeasured locations, and generate various RSC data and performance measures that can be used by maintenance personnel and road users.

The primary objective of this project is to continue the previous research efforts on developing highly transferrable and universally applicable methodologies, models, and tools for visualizing and inferring road surface conditions using data from RWIS and other road condition monitoring systems.

Ultimately, this project will provide winter maintenance personnel with newfound knowledge and analytical tools, of which they can employ to make better use of available resources, resulting in better maintenance and improvements to their highway infrastructure thereby promoting improved winter mobility and safety.

About the research

Department of transportation (DOT) maintenance supervisors utilize a variety of tools including maintenance decision support systems (MDSS) to gain a better understanding of current and future road surface conditions (RSC) during winter weather. MDSS automatically attempt to deduce current RSC based on road weather information system (RWIS) and other data to a greater or lesser degree of accuracy. Although current MDSS implementations present highway camera imagery, they typically do not incorporate automated camera image recognition in order to improve the MDSS assessment of winter RSC. Thus, there can be discrepancies between the road weather conditions in camera images and MDSS RSC assessments. For example, an MDSS assessment may determine that a highway is clear whereas associated camera images show snow or vice versa. Such discrepancies can lead to a loss of confidence, system criticism, and noncompliance with MDSS recommendations. From that point of view, the integration of automated RSC camera image recognition into MDSS implementations can have a number of benefits:

• Better RSC assessment performance
• Better road treatment recommendations owing to better RSC identification
• Improved MDSS use and compliance with system recommendations owing to user confidence

Recent research in RSC identification has applied convolutional neural networks (CNN) and related techniques to the winter RSC identification problem. The National Center for Atmospheric Research (NCAR) is interested in transitioning these automated RSC identification techniques from the research community to the DOT community. Since NCAR has significant experience in the research-to-operations arena, the NCAR team worked with Aurora Program members to develop a set of recommendations for transitioning the relevant technology to MDSS applications.

Even though recent research efforts seem quite successful, the NCAR team was also interested in the potential to improve the initial CNN RSC identification by incorporating additional relevant data such as the following:

• RWIS precipitation/temperature data
• Vehicle speed/volume data

About the research

The objectives of this project were to determine the relationship between weather conditions and roadway friction measurements as observed in the laboratory, determine whether it is possible to standardize friction measurements coming from multiple friction sensors for identical weather conditions and roadway pavement types, determine whether the relationship between weather and roadway friction found in the laboratory is analogous to the relationship between weather and pavement friction found in practice on highways, and model roadway friction using weather conditions to predict it at sites where friction measurements may not be available.

The objectives were accomplished through cold laboratory testing of stationary friction sensors, standardizing friction measurements from multiple stationary and mobile friction sensors, using meteorological measurements from Colorado and Minnesota to infer road friction conditions, and conducting a friction wheel measurement analysis using data from Sweden.

Key findings from this effort include the following:

• Machine learning models can be created using data from friction sensors in the cold laboratory that exhibit a good mean absolute error in predicting the laboratory friction response to meteorological conditions set in the laboratory, but the models have a higher mean absolute error when applied to data from the field. For this reason, the researchers do not recommend using the model developed in the laboratory with field data.
• Collocated road weather information system (RWIS) and stationary friction sensor data can be used to develop state-specific friction models using machine learning techniques. These models can then be used to provide a synthetic friction estimate at RWIS sites that are not equipped with stationary friction sensors. The accuracy of the predictions can be determined at sites where friction sensors are available. The accuracy is improved when water thickness and/or snow thickness are available.
• RWIS measurements including air temperature, surface temperature, dew point temperature, relative humidity, and road condition measurements including road state, water thickness, and snow thickness can be used to derive an accurate friction model that targets observed friction values.
• Friction values from multiple sensor types are close in magnitude when friction is high, but when friction values drop, agreement among sensors is variable. To standardize the measurements from multiple friction sensors, friction values from multiple sensors can either be averaged or associated with a set of friction categories.