Dr. Andreas Riener discussed the intersection of human factors and computer interaction in the evaluation of automated driving systems. This included simulations and the study of human-vehicle cooperation, emphasizing user experience.

Presented by Chantal Himmels, a Ph.D. candidate with BMW. The talk covered the role of driving simulators in automotive research, their validity, and their application in drawing real-world inferences.

Dr. Xuan (Sharon) Di discussed enhancing transportation digital twins through the 'brain'—utilizing AI to process sensor data for traffic optimization. She outlined applying physics-informed deep learning for traffic analysis and mean field games to model the interaction of autonomous and human-driven vehicles.

Prateek Bansal, Assistant Professor, National University of Singapore. Conventional population synthesis methods rely on household travel survey (HTS) data. However, the synthesized population suffers from a low spatial heterogeneity issue due to high data aggregation and low sampling rates of HTS data. Passively collected (PC) data from smartphone devices or transit smart cards have the potential to overcome the limitations of HTS data, thanks to the continuous collection of mobility patterns at a high spatial resolution for a large proportion of the population. However, the mismatched spatial resolution, sampling rate, and attribute information make the fusion of HTS and PC data challenging. This study presents a novel cluster-based data fusion method that exploits the benefits of both HTS and PC data to generate a synthetic population with high spatial heterogeneity. As the number of the value combinations for spatial attributes (e.g., home and work locations) in PC data is much larger than that in HTS data, clustering is adopted to deal with the high-dimensionality issue and link spatial attributes in the two data sources. The data fusion problem is then formulated as tractable multiple low-dimensional optimization subproblems. The properties of the proposed method are analytically derived. The application of the proposed approach is demonstrated using the HTS and LTE/5G cellular signaling data from Seoul, South Korea.

NYU's Xiyuan Ren, 4th year PHD candidate, and Prof. Joseph Chow.

One of the enduring challenges in statewide transportation planning is that consistent population travel data remains scarce, particularly for underserved and rural communities with inequity issues. This is changing with the availability of large-scale ICT data. A one-year project was initiated to develop a behavioral equity impact decision support tool based on NY statewide synthetic population data provided by Replica Inc. First, a NY statewide model choice model is developed to deterministically fit heterogeneous coefficients for trips along each census block-group OD pair, called group-level agent-based mixed (GLAM) logit. Considerations were made for four population segments, six trip modes, and twelve attributes. Second, a decision support tool for statewide mobility service region design was proposed. The tool is based on an assortment optimization problem with agent-specific coefficients and linear constraints, which can be efficiently solved through linear or quadratic programming (depending on variant). The decision support tool is applied to optimize service regions with one of the three objectives: (1) maximizing the total revenue; (2) maximizing the total change of consumer surplus; (3) minimizing the disparity between disadvantaged and non-disadvantaged communities.

Presented by UTEP Professors Dr. Kelvin Cheu and Dr. Ruimin Ke, and PHD Candidate Talha Azfar

A digital twin (DT) is a realistic digital model of a physical system or process, with sensors that measure real-world parameters and a simulation engine that replicates the behavior of the system or process. Realistic digital geographical models of real-world locations provide baseline information for digital twin applications. This project explores DT as a tool for the management of civil infrastructure. The real physical systems selected for this concept were a road network of the roundabout at a parking garage, a pedestrian bridge, and an interdisciplinary research building on the University of Texas, El Paso campus. Using the campus as a living lab, we developed DT models based on transportation networks, structural modeling, and LiDAR scans. The transportation network and 3D model of the campus were combined for traffic simulation and real-time sensing at a roundabout, while the digital model of a pedestrian bridge was made for structural simulations with provisions for strain and tilt sensors. Vehicles were detected and tracked from video on an edge computing device and visualized in the DT environment. DT models were analyzed considering various scenarios, demonstrating practical applications for real-world infrastructure. This presentation focuses on providing valuable insights into DT development, implementation challenges, and potential applications for civil infrastructure projects.

Presented by NYU's Hai Yang and University of Washington's Professor Jeff Ban

After five years of R&D, researchers have a number of independent simulation tools to evaluate different algorithms. A broad API was developed to handle interfacing any simulation with a multi-agent demand simulator. This was tested on the existing MATSim-NYC (which will be enhanced to include freight and parcel delivery activities) and aBEAM implementation, BEAM-NYC, for three use cases in electric transit, freight, and traffic. Each of these use cases considers equity impacts on different population segments (by income level, having disabilities, age level). The project jointly conducts some of the case studies in NYC and Seattle, enabling deeper insights of evaluated cases and promoting tech transfer and collaborations to broader communities (including agencies, the industry, and the public).

Presented by Professor Marcela A. Munizaga (University of Chile) and the Transportation Research Board Subcommittee AEP30(2)/AEP40 on Route Choice and Spatio-Temporal Behavior

Data automatically generated by the technological devices used in the operation of public transport systems provide a unique opportunity to observe travelers' behavior. These big databases allow monitoring the system from different perspectives, such as bus operations, level of service, demand, across time and space. We have worked on all of those aspects. This presentation will focus on work devoted to improving the understanding of route choice behavior among public transport passengers. By utilizing revealed preference data from passive transportation sources, we developed methodologies to model the heterogeneity in passengers' route choices, the process of generating consideration sets in intermodal route choices, and the passengers' learning process when facing new route alternatives. These methodologies were applied to the case of Santiago, Chile, where we found evidence supporting the need to adapt traditional modeling approaches, commonly used with traditional transportation data, to a modeling context with massive and disaggregated data derived from widely applied transportation technologies.

Presented by: Haggai Davis, III, 3rd year PHD candidate at NYU and Joseph Chow, Associate Professor at NYU and Deputy Director of C2SMART

Due to its complexities, urban freight modeling requires robust data inputs that must be combined from numerous sources which can often limit the accuracy, power, or transferability of a model. Even with proper inputs, new techniques must be developed to make use of all the information that the data contain while still being feasible in their implementation. This presentation addresses the challenges in turn by estimating FTG from public data sources, inputting the FTG into a tour creation model, and then solving an entropy maximization problem with an iterative balancing algorithm to load flows onto the tours. New York City is then used as an example to illustrate and validate the application of these techniques.

Presented by Prof. Kaan Ozbay, NYU, C2SMART

Proper calibration process is key for traffic safety evaluations using simulation models. Allowing for a with and without comparison under controlled environment that is not directly testable in the field, microsimulation-based approach has drawn considerable attention for the performance evaluation of emerging technologies, including connected vehicle (CV) safety applications. Different from the traditional approaches to evaluate mobility impacts, safety evaluations of such applications demand the simulation models to be well calibrated to match real-world safety conditions. This seminar will present a novel calibration framework which combines traffic conflict techniques and multi-objective stochastic optimization to calibrate the operational and safety measures simultaneously. The conflict distribution of different severity levels categorized by time-to-collision (TTC) is applied as the safety performance measure. Simultaneous perturbation stochastic approximation (SPSA) algorithm, which can efficiently approximate the gradient of the multi-objective stochastic loss function, is used for model parameters optimization that minimizes the total simulation error of both operational and safety performance measures. A case study will be demonstrated by calibrating a microscopic simulation model to evaluate CV safety applications as a part of the NYC Connected Vehicle Pilot Deployment Program.

Presented by Dr. Wen-Long Jin, UC Irvine

In this talk, Dr. Wen-Long Jin will present some of his recent results on new approaches and paradigms in traffic flow modeling and control. He will first discuss traffic flow models in three types of spaces: (1) provably safe driving models for both human-driven and autonomous vehicles in the absolute space on a road, (2) bathtub models for network trip flows in a relative space with respect to individual travelers’ remaining trip distances, and (3) day-to-day traffic flow models for departure time choice in an economic space with respect to the scheduling cost. Then he will present two studies on traffic system operations and control: (1) dynamic pricing schemes for high-occupancy-toll lanes with a single or multiple bottlenecks; and (2) fleet-size management for shared mobility systems with for-hire vehicles.

Presented by Dr. Carmelo D’Agostino, Lund University (Sweden)

Traffic accidents are among the leading causes of death for people aged 5–35 worldwide, causing transport externalities and thus unsustainability. At the same time, the introduction of almost fully connected and automated vehicles (CAVs) (levels 3–4 of SAE) is already a reality. Although CAVs go in the direction of smart mobility, their sustainability is still questionable because their deployment in open traffic introduces unexplored risks. Indeed, while technological progress is rapidly being pursued, there remain significant issues related to the development and integration of CAVs with physical and digital infrastructure and to their user acceptance on shared roads. The main reason is a general perception that they are not safe and thus may introduce inequality. In this context, neither the actual accident-based nor proactive methods for road safety analysis can be applied when CAVs interact with conventional users. This is primarily due to the lack of knowledge about the influence of the digital and physical infrastructure in the interactions among vehicles in mixed traffic conditions. In this framework, the use of simulation and virtual reality, combined with validation on real world scale, represents the only approach to provide the basis for new computational methods for infrastructure safety assessment in future mobility scenarios based on a rigorous scientific approach. The use of simulation at different levels combined with new Surrogate Measures of Safety (SMoS) can address the problem of the safety evaluation of the interaction between conventional vehicles and CAVs. The seminar will present how virtual reality and simulation are being used as a tool to replace naturalistic observations in the real-world and their pros and cons, in three different research projects on CAVs safety among which is a European Research Council Grant.

Presented by UW’s Prof Jeff Ban and Ohay Angra

In this seminar, we will talk about the motivation, design framework, and initial implementation of a multi-scale transportation simulation platform. The platform simulates the movements of connected/automated vehicles (Unity), traffic flow dynamics (SUMO), and demand generation on a large-scale transportation network (MATSim). We focus on the SUMO-MATSim development details and the current challenges of integrating the different simulation tools. We will also mention how the multi-scale simulation platform may be used for conducting research for emerging technologies and systems in transportation.