Principal Investigator: Li Jin, NYU

This project aims to develop a systematic way to design smart highway systems with networked video monitoring and control resiliency against environment disruptions and sensor failures.

On the video monitoring side, we will investigate 1) efficient deep learning methods for extracting fine-grained local categorical traffic information from individual surveillance videos (e.g., traffic mixture, environment information, anomaly/extreme-weather detection in the scene), and 2) novel graph neural network (GNN) methods to correlate and propagate the local information through the highway network for global states estimation (e.g., vehicle tracking and reidentification, traffic prediction in unobserved area). We will discretize the traffic volume of a local road intersection into several levels/categories (e.g., low, mid, high), which will make our method less prone to estimation errors that could originate from challenging environmental variations.

On the system design side, we will 1) establish dynamic models for capacity using video data, 2) model failure in either cyber or physical components, 3) study the relation between sensor deployment and observability for resilient traffic control (e.g. route guidance and ramp metering).

The expected outcome is an implementable approach to designing resilient smart highway systems with trustworthy monitoring capability. We also expect our approach (with appropriate modification) to be applicable to general transportation systems.

Principal Investigator: Quanyan Zhu, NYU

Our research aims at developing a resilient framework to be applied to transportation systems using connected and autonomous vehicles (CAVs). This responds to vehicular accident rates often related to inefficient communication systems. A Vehicular Ad hoc Network (VANET) is a significant innovation toward avoiding such deadly traffic mishaps with the assistance of a variety of state-of-the-art safety applications. A VANET is a self-organized, multi-purpose, service-oriented communication network enabling vehicle-to-vehicle and vehicle-to-roadside infrastructure communication for the purpose of exchanging messages to ensure an efficient and comfortable traffic system on roads. Its value, however, can potentially be impaired by cyberattacks. In particular, the focus of this research will be on false data injection attacks, in which a malicious agent aims at affecting CAV behavior by injecting in the network false information concerning, for example, the traffic condition in the area or the availability of charging stations for the CAVs. The countermeasures will be developed using anomaly identification techniques based on learning and detection algorithms. In particular, the data collected from the nearby CAVs will be utilized to simulate and estimate the time evolution of the system; based on the outcome of such simulation, the CAV will apply a decision logic to distinguish between useful information and malicious data. In summary the approach aims at developing security game frameworks for vehicular networks, that model the interaction between malicious attackers to VANETs and various defense mechanisms protecting them. Mitigation measures will be applied emphasizing multi-modal connectivity involving CAVs and/or traditional vehicles. This connectivity introduces flexibility in transport options. Such flexibility also captures similar concepts for communication and information technologies such as redundant sensors and various kinds of backup systems and countermeasures. This research is designed to provide guidance for users and developers of safe vehicular systems. The team’s proposed work will support the goal of C2SMART for research in disruptive technologies, engagement and training and education. The team will engage more with the researchers at the Center, building upon and expanding the team’s familiarity with and in some cases work experience with many of the Center faculty and researchers. Technology transfer is an important part of the research and outreach efforts. We plan to generate data sets and develop simulation tools and research results for the research community in a usable form. The location of data sources upon which the research is based and research results will be placed in a data repository. As part of the outreach effort we will a prepare a guide and the outline for a demonstration for users to use the data.

Principal Investigator: Semiha Ergan, NYU

Despite increased regulations, restrictive measures, and devices used for warning, work zone injuries and fatalities are still observed at highway construction projects with alarms/notifications being ignored. With a vision to reduce the number of injuries and fatalities, this project aims to understand the key parameters (e.g., work zone location characteristics, personal vigilance levels, types of construction work) that play roles in achieving responsive behaviors in workers. Key questions this research answers are, at what conditions people ignore/response to warnings at work zones? How we can calibrate notification systems for getting responsive actions from workers? What are the modalities, frequencies, and timings of pushing notifications in these calibrated systems? Through wearable sensors and realistic representations of work zones in virtual reality, we plan to collect worker behavioral and physiological (heart rate) responses to alarms/warnings/notifications issued under various realistic scenarios and modalities of warning mechanisms (e.g., sensory, visual, audial). With a reinforcement learning based approach, the collected data will be used for determining expected worker/driver behaviors (validated through subjects’ heart rate data) when prompted with an alarm/warning/notification learned from similar behaviors. The outcome of this research will help to calibrate when, at what frequency, and how to (with what modalities) share warnings with habitants of work zones for effective responses towards reduction of incidents.

Principal Investigator: Joseph Chow, NYU

In Year 2 we developed the foundations of a virtual test bed for NYC using large-scale transportation simulation models built with MATSim and SUMO. We propose to continue this research in Year 3 in further expanding the capabilities of the virtual test bed, applying the models to scenarios of interest to NYC DOT and the MTA, and focusing on tech transfer activities to make the ecosystem accessible to policymakers and our consortium partners. Applications include on-demand robotic taxi, traffic flow modifications to allow for connected vehicles, and dockless bikeshare.

The research will provide resources and a virtual framework for supporting and helping public sector’s decision making to fill in the gap between basic research and field deployment. Subsequent implementations of the test bed in other cities will form the basis for a “Network of Living Labs” that NYC DOT and other local agencies can benefit from shared knowledge transfer.

Principal Investigator: Joseph Chow, NYU

As an extension of the MATSim virtual test bed in development under Year 2, we plan to apply it to the evaluation of the Brooklyn bus network redesign proposed by Eric Goldwyn and colleagues at Marron Institute. We will pay specific attention to the presence of ride hail fleets (through analytical models and exploratory consideration of MATSim extensions) in justifying stop spacing and route allocations.

The project serves as a demonstration of evaluation of bus redesign in the presence of ride hail using analytical and simulation tools. Lessons learned from this will be of great value to many cities plagued with resource constraints in operating buses and having companies operating ride hail services.

Principal Investigator: Stanislav Sobolevsky, NYU

The project will develop a citywide data-driven transportation simulation modeling framework for probabilistic assessment of the associated mode-shift and resulting environmental, social and economic impacts of ride-sharing solutions (e.g. UberPOOL, Lyft shared etc) on urban transportation system in New York City efficiently leveraging available partial transportation data. The impacts in question include: travel time cut for passengers, reduction of traffic, gas consumption/ emissions by type (CO, NOx, PM2.5), travel time/cost savings for passengers, increased earnings for Lyft and Uber drivers, jobs for for-hire-vehicle drivers. Once developed, the new framework is readily applicable to the predictive assessment of the impacts of many other transportation pricing and policy decisions, such as Manhattan congestion charge which is scheduled to come into effect by January,2019 – additional use case depicting it will be provided.

Principal Investigator: Don Mackenzie, University of Washington

This project will build on prior C2SMART work to develop a tool for calculating the expected profit impacts of relocating vehicles in a free-float carsharing system. Key innovations are: (1) combining macro-scale and micro-scale relocations, and (2) considering expected profit over multiple subsequent trips, rather than just the next trip. Methods resulting from this work can increase the efficiency of use of carsharing vehicles, and increase profitability of operating a carsharing system.

Principal Investigator: Jeff Ban, University of Washington

While connected and automated vehicles (CAVs) have received much attention in transportation especially on how they may transform future urban traffic/vehicle control, it is becoming increasingly clear that in the near future, we will have to deal with a relatively low penetration of CAVs with limited level of autonomy (e.g., Level 2 or Level 3). Thus how to understand and test in real world the benefits of CAVs with limited penetration and autonomy for vehicle-traffic control remains an interesting and imperative question. This research aims to extend and field test the CAV-based traffic signal/vehicle control methods the team has developed in the past (many were supported by C2SMART) to understand and quantify the benefits of CAV-based control in real world. The team will collaborate with UW’s EcoCar3 team to conduct field testing and to involve students from different levels (Ph.D., graduate and undergraduate students) and backgrounds to participate in this multidisciplinary and cutting edge research project.

Principal Investigator: Kelvin Cheu, University of Texas at El Paso

The University of Texas at El Paso (UTEP) and New York University (NYU) are proposing a continuation of the projects carried out during Years 1 and 2 of this UTC.  In the past 2 years, our research team conducted initial surveys in El Paso and in New York City to understand the lifestyles and mobility needs of the senior adults. A smartphone application prototype named Urban Connector (UC) was developed to cater to the seniors in El Paso. A follow-up survey was conducted to gather feedbacks on the UC application prototype. Subsequently the UC prototype was improved to its beta version.

In Year 3 of this C2SMART project, the objectives are:

1. To customize the beta version of the UC application for New York City;

2. To perform a beta test in New York City and gather user feedbacks.

3. To fine-tune the UC application, based on the information collected in El Paso in Year 2 (the third survey), to release Version 1.0.

4. To collect a set of UC application usage data through an anonymous pilot test;

5. To explore a data analysis framework that will characterize a UC application user’s mobility patterns without knowing the person’s identification.

Principal Investigator: Kelvin Cheu, University of Texas at El Paso

Most universities are constantly challenged by the problem of inadequate parking supply to meet the demand. The common parking management policies are zoning of parking lots (allocation of stalls to different types of users), differentiation of permits and pricing. Every university has a parking department responsible for the implementation of parking policies.  We refer to this department as the University Parking Office (UPO). As an initial step towards finding a comprehensive parking solution, this research focuses on student parking, which is the largest group of parking users on campus.

The objectives are:

1. To conduct a survey to understand the factors that influence students’ parking location choices, usage patterns, preferences among the different Intelligent Transportation Systems (ITS) applications for parking and levels of tolerance for parking search time.

2. To develop the level of service (LOS) criteria for parking search time;

3. To develop a student parking lot zoning and zone permit pricing (Z2P2) model that groups several parking lots into a zone, and recommend the permit price for the zone;

4. To develop Version 1.0 of a software tool named Sparkman (acronym for Smart Parking Management) that estimates a campus’ total student parking demand, the “base price” of student parking permits, zoning of student parking lots, and the permit prices of the different zones.  The total demand and “base price” models have been developed by the PI as two separate models in earlier research, while the Z2P2 models will be developed as part of this project (objective 1). The total demand, “base price” and Z2P2 models will be integrated into Sparkman as part of this research.

The scope of this project does not include faculty and staff parking, parking for carpool, electric, university, service, freight vehicles, vehicles with disabled license plates, design of campus transit and bike sharing systems, dynamic parking pricing and app-based stall reservation system. These may be addressed in future versions of Sparkman.

Principal Investigator: Camille Kamga, City College of New York

New York City operates a large and diverse fleet from sanitation trucks to light‐duty vehicles. Each type of vehicle has it’s own safety needs. More than 50% of crashes involving city vehicles could have been avoided. 93% of vehicular crashes, in general, involve human errors according to NHTSA. Connected vehicle technology is being deployed in over 8,000 vehicles in NYC as a part of the CV pilot. About 500 of these vehicles belong to the city municipal fleet.
The goal of this study is to explore aspects of connected vehicle technologies that can be implemented for municipal fleet in New York City. The CV benefits intended to be realized in the NYC CV pilots are vehicular and pedestrian safety. CV apps included in the NYC CV pilot will tested against the crash types and potentially dangerous scenarios involved by NYC fleet vehicles. These apps will be evaluated for their effectiveness using the NYC CV pilot simulation testbed built as a part of the previous study during years 1 and 2. Various benefits safety, environmental and financial for these apps will be projected to scale up to the fleet along other long‐term considerations including potential alternatives and recommendations on adoption of technologies will be made. The benefits quantification can help NYC fleet in expanding their CV deployment and also other agencies in their adoption.

Principal Investigator: Hani Nassif, Rutgers University

This project will evaluate the damage cost associated with the overweight trucks that violate the federal weight regulation from a national perspective using national WIM data from LTPP, SHRP, etc.  Moreover, Advanced Weigh-In-Motion (A-WIM) system with high accuracy sensors will be deployed to develop an autonomous enforcement approach to minimize the infrastructure damage. The outcome of this project will be damage cost models for bridges and pavement from a national perspective.  In addition, provisional guidelines for enforcement using A-WIM technology will be documented to minimize the damage associated with the trucks against the federal truck weight regulation.