This research aims to explore the possibility of using sec-by-sec traffic signal data provided by the Centracs system to evaluate intersection safety. Intersection safety has long become a national concern. However, traditional methods, either using historical crash data collected from infrequently happened collisions, or potential conflicts estimated from microscopic simulation which assumes “accident-free”, cannot provide accurate and timely evaluation of intersection safety. By contrast, this research estimates potential traffic conflicts using sec-by-sec data extracted from Centracs, therefore has the potential to provide timely and accurate intersection safety information. Centracs has gained more and more popularity in the nation, especially in California, due to its robust capability for improving intersection efficiency. Especially, Centracs collects and archives sec-by-sec traffic and signal data, which can be used to evaluate safety. Using sec-by-sec data, this research proposes an innovative method to estimate potential traffic conflicts and predict red-light violations, which essentially indicate the safety level of intersections. The proposed research will be tested using the data collected from 5 intersections in Anaheim, CA. This research is expected to contribute significantly to the improvement of intersection safety, and build a foundation for future dynamic systems that could alert drivers of emerging and impending hazardous situations.

In this research, we propose a unified framework for (1) analyzing dynamical and stationary patterns subject to different control strategies; and (2) designing control strategies based on understanding of traffic patterns. We will describe the evolution of traffic dynamics in a signalized network by the Link Transmission Model (Yperman et al., 2006; Yperman, 2007), which, together with Newell’s simplified kinematic wave model (Newell, 1993), is another formulation of the network kinematic wave theory based on the LWR model. In (Jin, 2014), two continuous formulations of LTM were derived from the Hopf-Lax formula for the Hamilton- Jacobi equation of the LWR model. Then we will (1) analytically derive macroscopic fundamental diagrams (MFD) for stationary traffic patterns with different network topologies, road conditions, driving behaviors, and signal settings; (2) quantify congestion mitigation effects of different signal settings, including cycle lengths, green splits,  and offsets, as well as speed limits and road lengths; (3) formulate an optimization problem to find optimal road, speed limit, and signal control parameters under certain demand levels, and (4) develop a set of simple decision-support tools for arterial network improvement.

The California Department of Transportation (Caltrans) along with other State Departments of Transportation (DOTs) are required to submit axle-based classification count reports to the Federal Highway Administration (FHWA) under the Highway Performance Measurement System (HPMS) program.  The data are an input to the federal government’s allocation of funds to the states to effectively maintain pavement quality in high priority corridors.  However, safety concerns and high costs are associated with existing methods of data collection.  For example, setting up road tubes and other temporary data collection devices frequently exposes DOT personnel to safety hazards due to their close proximity to the traveled lanes of high speed corridors.  In addition, permanent detection systems such as existing piezo-based vehicle classifiers are expensive to install and are associated with high maintenance costs due to their frequent failures.
The statewide Truck Activity Monitoring System (TAMS) developed by UC Irvine currently provides continuous truck counts by vocation at 70 major truck corridors in California, and will be expanded to over 90 locations by the end of 2016.  The system is based on relatively inexpensive updating of hardware in roadside cabinets at existing traffic detection sites that equip existing permanent loop sensors with inductive signature technology.  Although the current classification models in TAMS are focused on truck vocations, initial investigations have shown that there is excellent potential to successfully develop truck classification models that are capable of classifying vehicles according to the axle-based FHWA HPMS scheme using only inductive loop signature data.  In this proposed study, we will partner with advisors from the Caltrans Traffic Census Program (TCP) to develop HPMS-based classification models, and design a streamlined solution within TAMS to process the data for HPMS reporting requirements. Existing data as well new data sources collected at WIM sites will be used to develop and validate the models, while TCP advisors will provide reporting guidelines to ensure that the developed system meets stakeholder needs.  We will also investigate establishing a test detection site using only solar power to study the feasibility of implementing this solution at off-grid sites to further extend the applications of this research.