Location-based social networks (LBSN) are social media sites where users check-in at venues and share content linked to their geo-locations. LBSN, considered to be a novel data source, contain valuable information for urban planners and researchers. While earlier research efforts focused either on disaggregate patterns or aggregate analysis of social and temporal attributes, no attempt has been made to relate the data to transportation planning outcomes. To that extent, the current study employs LBSN service-based data for an aggregate-level transportation planning exercise by developing land-use planning models. Specifically, we employ check-in data aggregated at the census tract level to develop a quantitative model for activity intensity as a function of land use and built-environment attributes for the New York City (NYC) region. A statistical exercise based on clustering of census tracts and negative binomial regression analyses are adopted to analyze the aggregated data. We demonstrate the implications of the estimated models by presenting the spatial aggregation profiling based on the model estimates. The findings provide insights on relative differences of activity engagements across the urban region. The proposed approach thus provides a complementary analysis tool to traditional transportation planning exercises.

Research indicates that the projection of traffic volumes is a valuable tool for traffic management. However, few studies have examined the application of a universal automated framework for car traffic volume prediction. Within this limited literature, studies using broad data sets and inclusive predictors have been inadequate; such works have not incorporated a comprehensive set of linear and nonlinear algorithms utilizing a robust cross-validation approach. The proposed model pipeline introduced in this study automatically identifies the most appropriate feature-selection method and modeling approach to reduce the mean absolute percentage error. We utilized hyperparameter optimization to generate a universal automated framework, distinct from model optimization techniques that rely on a single case study. The resulting model can be independently customized to any respective project. Automating much of this process minimizes the work and expertise required for traffic count forecasting. To test the applicability of our models, we used Florida historical traffic data from between 2001 and 2017. The results confirmed that nonlinear models outperformed linear models in predicting passenger vehicles’ monthly traffic volumes in this specific case study. By employing the framework developed in this study, transportation planners could identify the critical links on US roads that incur overcapacity issues.

Vehicle-to-infrastructure (V2I) communication is essential for reliable deployment of connected automated vehicle technology, contributing to the advanced safety and optimization of our transportation networks. However, supplying and maintaining necessary wireless infrastructure is a challenging task, particularly when it comes to rural areas. This study proposes a novel methodology that uses artificial intelligence, machine vision, and smart traffic signs to support V2I in areas where availability of wireless communication infrastructure is limited. The objective of this paper is to investigate the operational challenges of the proposed low-cost solution in different V2I applications, including a MapData message in an unsignalized traffic intersection, traveler information message in a work zone, and a red-light violation warning with the help of a smart sign. The proposed system showed some important advantages, such as invulnerability to third-party alterations and robust operation under harsh environmental conditions.

Adversarial attacks pose serious traffic safety risks that could potentially impede the deployment of automated driving systems (ADS). Recent discoveries show that imperceptible perturbations (also referred to as adversarial attacks) to inputs of deep neural networks employed in these systems could impair their image recognition and prediction capabilities. Adversarial attacks can misguide advanced driving assistance systems (ADAS) to misclassify a traffic sign (e.g., classify a stop sign as a speed limit sign), causing the vehicle to accelerate instead of stop, which may potentially result in catastrophic incidents. This paper investigates encrypted invisible security patches placed on traffic signs and an onboard machine-vision system that can be integrated into the ADAS (i.e., ADS and cooperative adaptive cruise control systems). The security patches are essentially a visual representation of encrypted hash values generated via a one-way cryptographic algorithm. In this way, every traffic sign will be associated with a unique security pattern that can only be recognized by the proposed machine-vision technology. Additionally, these patches will be made invisible with the help of ultraviolet reflective sign technology. The proposed methodology aims to bring an additional security layer to sign recognition to enhance the reliability of ADAS. If a traffic sign is physically altered to attempt an adversarial threat, the security patch on the sign will not match with the sign information, and the driver will receive a warning to take control of the driving. For self-driving vehicles, a signal will be sent to the autonomous vehicle to enforce a safe driving action.

The impact of freight on the transportation system is accentuated by the fact that trucks consume a greater roadway capacity than other vehicles and therefore cause more significant problems including traffic congestion, traffic delays, crashes, and pavement damage. Evaluating the actual repercussions of truck traffic becomes paramount in locales where roadway expansion is unfeasible. Trucks are vital to the economy, providing essential services to commerce and industry, and yet it is crucial that their operation does not contribute to the deterioration of infrastructural quality or compromise public safety. Currently, we lack methodologies in practice for the real-time management of traffic, specifically for truck routing, to minimize travel times and prevent delays due to non-recurrent congestion, such as traffic incidents. Accordingly, this study aimed to devise a truck routing strategy utilizing a traffic micro-simulation model (VISSIM) and to assess its effects on reducing travel delays. This involved the development of real-time truck re-routing simulation models that take into account non-recurrent congestion and the resulting travel delays and fuel consumption. The VISSIM model was applied to the I-75 corridor in Marion County, Florida, focusing on non-recurrent congestion effects on travel delays and fuel consumption. The initial findings suggest that the implementation of a dynamic truck re-routing system can significantly alleviate traffic congestion, resulting in a marked decrease in travel delays and fuel usage, demonstrating the potential for such strategies to enhance the overall efficiency of the transportation system.