2017 Transportation Research Board Annual Meeting

Please join us!

TransInfo will again be hosting a reception at TRB this year. Please join us on Monday, January 9th from 7:30 - 9:30 PM in the Judiciary Square Room at the Marriott Marquis Hotel. We hope to see you there for food, drinks and networking! 

ISTL Faculty and Students: Posters, Papers and Presentations

Check out all of the sessions University at Buffalo faculty and students are participating in below. 

Monday, January 9th, 2017

8:00 AM- 9:45 AM
Poster Program: Urban Freight Innovations

Crowdsourcing Last-Mile Delivery of Online Orders by Exploiting Social Networks of Retail Store Customers

Aashwinikumar Devari, University at Buffalo
Alexander Nikolaev, University at Buffalo
Qing He, University at Buffalo

Abstract: Most major retailers and organizations strive to provide speedy and efficient delivery of products and explore the opportunities for saving on their last mile delivery costs. Crowd logistics is a subject of high interest in such endeavors. However, at its current state of development and adoption, further research is required to control and improve upon crowdsourced delivery times, risks and costs. This paper demonstrates the potential benefits of crowdsourcing delivery operations exploiting a social network of the customers of a retail store in assisting with the last mile delivery. In this paper, we conceive of a social network that connects the customers who are co-workers and/or neighbors of each other. The presented models and analyzes are informed by the results of a survey conducted with 101 participants to gauge people’s attitudes towards package delivery to and by friends or acquaintances. Relying on the survey responses, a logistic regression model is built to predict the probability of a package being delivered from a store to a customer by the customer’s friends. In order to study a potential large-scale impact of such delivery mechanism, we set up a simulation environment in TRANSIMS, an activity-based transportation modeling tool with the data collected from a real-world city. The results of the simulated experiments indicate that, by exploiting crowdsourcing, a retailer in a small city can reduce truck mileage by 57%, which is equivalent to reducing delivery costs by 8600USD per day. On average, each delivery adds extra 10 minutes to the regular trip of the party providing the delivery assistance. As a result of this assistance, the expected achieved reduction in pollutants, i.e., NOx, PM and PM, emitted by delivery trucks amounts to nearly 55%.

              

8:00 AM- 9:45 AM

Lecture Program: Social Media and Travel: Explorations Using New Data and Methods

Exploring Travel Behavior with Social Media: An Empirical Study of Abnormal Movements Using High-Resolution Tweet Trajectory Data

Zhenhua Zhang, State University of New York (SUNY)
Qing He, University at Buffalo
Shanjiang Zhu, George Mason University

 Abstract: This study reveals the characteristics of travel behavior using high-resolution Twitter data through a series of empirical studies and further explains the abnormal movements by the tweet trajectories. First, this paper explores the characteristics of individual travel behavior especially the location geo-distribution, movement scale and the clustering features of undirected travel. Second, this paper proposes a geo-mobility clustering method that groups the tweet locations driven by the same travel motif. This clustering method captures the clustering features of traveler’s hourly locations and detects the abnormal travel behavior. Third, the tweet posts are examined to identify the social activities behind these abnormal movements. The results of our algorithm shows that 46.2% of the abnormal movements can be tied with social activities by the keywords of the tweets.

 

10:15 AM- 12:00 PM

Poster Program: Statistical Methods in Transportation

Panagiotis Anastasopoulos, University at Buffalo, presiding

 

10:15 AM- 12:00 PM

Lecture Program: Railroad Track Performance Metrics

Data-Driven Optimization of Railway Track Inspection and Maintenance Using Markov Decision Process

Siddhartha Sharma, University at Buffalo
Yu Cui, University at Buffalo
Qing He, University at Buffalo

Abstract: Railway big data technologies are transforming the existing track inspection and maintenance policy deployed for Class I railroad in North America. This paper develops a data-driven condition-based policy for the track geometry inspection and maintenance. Both preventive maintenance and corrective maintenance are taken into account by the investigation on a 33-month inspection dataset which contains a variety of geometry measurements for every foot of track. First, this study separates the data based on the time interval of inspection run, calculates the aggregate TQI for each track section, and predicts the track spot geo-defect arrival probability with Random Forests. Then a Markov Chain is built for modeling aggregated track deterioration, while the spot geo-defect is modeled by a Bernoulli process. Finally, a Markov Decision Process (MDP) is developed for track maintenance decision making and optimized by using value iteration algorithm. By comparing with existing maintenance policy with Markov Chain Monte Carlo (MCMC) simulation, the new maintenance policy developed in this paper results in a saving around 10% of total maintenance costs for every 1 mile of track.

 

10:15 AM- 12:00 PM

Poster Program: Multimodal Operations at Signalized Intersections

Multimodal Hierarchically Responsive Signal Control with Lexicographical Dynamic Programming Approach

Qing He, University at Buffalo
Hernan Caceres Venegas, Universidad Catolica del Norte (Chile)
Manoj Reddy Kandukuri, University at Buffalo
Zhenhua Zhang, State University of New York (SUNY)

Abstract: This paper develops a Multi-modal Hierarchically Responsive Signal control model called MARS for trajectory-based signal control, by assuming that high penetration of floating sensors (e.g. Connected Vehicles, Smartphones, etc.) is available. First, this study conducts a comprehensive survey with traffic signal professionals, who bring up existing state-of-practice, open issues and future challenges in multi-modal traffic signal control. This survey also identifies the issues of current weight-based modeling for multi-modal control. It is found that assigning weights cannot be tied with delay of each mode in a straight forward manner. Second, by using multi-modal trajectory data, this paper develops a hierarchically multi-modal signal control model, in which each travel mode is solved by a dynamic programming hierarchically with the consideration of the delay and budget from upper-level modes. Further, the proposed control model is evaluated by microscopic simulation tool VISSIM at an isolated intersection, including three competing travel modes: light rail, buses and passenger cars (with trucks).

 

1:30 PM- 3:15 PM

Poster Program: New Research on Travel Time, Speed, and Reliability Data

Analyzing Travel Time Reliability and Its Influential Factors on Emergency Vehicles with Generalized Extreme Value Theory

Zhenhua Zhang, State University of New York (SUNY)
Qing He, University at Buffalo
JIZHAN GOU
Xiaoling Li, Virginia Department of Transportation

Abstract: Travel time reliability is very critical for emergency vehicle (EV) service and operation. The travel time characteristics of EVs are quite different from those of ordinary vehicles (OVs). Although EVs own highest road privilege, they may still experience unexpected delay that results in massive loss to the society. In this study, we employ the generalized extreme value (GEV) theory to measure and analyze extremely prolonged travel time. Among three GEV distributions, Weibull distributions are found to be the best distribution model according to several goodness-of-fit test. A new reliability index is derived to measure travel time reliability. Numerical examples demonstrate the advantages of GEV-based reliability index over variance and percentile value in the applications of EV. We further investigate the potential influential factors on EV travel time reliability. Results show that link length and number of lanes may have a negative impact on the link reliability. To some extent, left-turn traffic volume, and the number of left-turn lanes, also have negative impacts on travel time reliability.

 

1:30 PM- 3:15 PM
Poster Program: User Perspectives and Impact Assessments of Vehicle-Highway Automation Systems

Public Acceptance of Driver-Assist and Warning Technologies

Andrew Bartlett, University at Buffalo
Adel Sadek, University at Buffalo

Abstract: The last decade has seen a surge in the implementation of new onboard technologies in personal vehicles. While the emergence of fully autonomous vehicles (AVs) on the market may occur in the near future, the use of lower levels of automation in vehicles, such as driver-assist systems (e.g. self-parking, adaptive cruise control, collision avoidance) and driver warning systems (e.g. collision avoidance, blind spot, lane departure, and congestion/incident warnings) already exist and are becoming more prominent. The primary objective of this study was to determine what factors make individuals more or less comfortable with using or owning vehicles which incorporate a variety of driver-assist and warning technologies. Survey results revealed that a small majority said they would be comfortable with driver-assist technologies, such as self-parking (57%), adaptive cruise control (61%), and collision avoidance (58%). A much larger majority felt comfortable with warning technologies, including warnings for congestion/incidents (87%), lane departures (81%), blind spot occupancy (89%), and collisions (81%). A series of well-fitting models led to the identification of a variety of relationships between demographic information, driving behavior, and opinions about driving technologies. Those who stated they understood the benefits of AVs were found to be more comfortable with all driver-assist and warning technologies across all models. In general, those who were older, had higher incomes, and had higher levels of education were more comfortable with the technologies described. Those who had recently been ticketed for a driving violation were less comfortable with driver-assist technologies but this was not seen to impact comfort with warning technologies. Drivers with longer commutes were less comfortable with some warnings, but more comfortable with self-parking. These results could be applied to help shape the marketing of these technologies to certain demographics, as well as aid in forming public policy related to their implementation.

3:45 PM- 5:30 PM

Poster Program: Driver Behavior and Inattention

Exploratory Empirical Analysis of Measured and Perceived Aggressive Driving Behavior in a Driving Simulation Environment

Nima Golshani, University of Illinois, Chicago
Md Tawfiq Sarwar, NRC Research Associateship
Panagiotis Anastasopoulos, University at Buffalo
Kevin Hulme

 Abstract: In this paper, driving simulation data and surveys collected in the spring of 2014 in Buffalo, NY, are used to explore the factors that can affect measured (through driving simulation experiments) and perceived (self-reported, based on surveys) aggressive driving behavior.  To simultaneously account for unobserved heterogeneity, panel data effects, and cross equation error correlation, a random parameters with within-panel heterogeneity bivariate probit model is estimated.  The results show that a number of factors affect the measured and perceived aggressive driving behavior, such as: driving experience and exposure (frequency and willingness to drive, accident history, and driver experience); socio-demographic characteristics (family status, household income, level of education, and grow up area), and behavioral and other characteristics (traffic violation warning, music preference, and alcohol consumption tendency).  The results imply that some drivers may drive aggressively when they perceive their driving behavior as non-aggressive (or the opposite), and that different factors play in how aggressive driving behavior is measured and perceived.

Tuesday, January 10th, 2017

8:00 AM- 12:00 PM

Program: Artificial Intelligence and Advanced Computing Applications Committee

Sherif Ishak, Louisiana State University, presiding
Adel Sadek, University at Buffalo, presiding

8:00 AM- 12:00 PM

Poster Program: Crash-Based Research: Wrong-Way Driving, Work Zones, Debris, Taxis, and Other Crash Situations

Analysis of Stationary and Dynamic Factors Affecting Highway Accident Occurrence

Grigorios Fountas, University at Buffalo
Md Tawfiq Sarwar, NRC Research Associateship
Panagiotis Anastasopoulos, University at Buffalo
Alan Blatt, CUBRC
Kevin Majka, CUBRC

Abstract: Traditional accident analysis typically explores non-time-varying (stationary) factors that affect accident occurrence on roadway segments.  However, the impact of time-varying (dynamic) factors is not thoroughly investigated.  This paper seeks to simultaneously identify pre-crash stationary and dynamic factors of accident occurrence, while accounting for unobserved heterogeneity.  Using highly disaggregate information for the potential dynamic factors, and aggregate data for the traditional stationary elements, a dynamic binary random parameters (mixed) logit approach is adopted.  With this approach, the dynamic nature of weather-related, and driving- and pavement-condition information is jointly investigated with traditional roadway geometrics and traffic characteristics.  The analysis is based on crash and non-crash observations between 2011 and 2013, drawn from urban and rural highway segments in the state of Washington.  The findings show that the proposed methodological framework can account for both stationary and dynamic factors affecting accident occurrence probabilities, and for unobserved heterogeneity.  The results also demonstrate the potential of random parameters modeling, in terms of forecasting accuracy and explanatory power. 

10:15 AM- 12:00 PM

Poster Program: Current Issues in Alternative Transportation Fuels and Technologies

Incorporating Demand Dynamics in Multi-period Capacitated Recharging Location Planning for Electric Vehicles

Anpeng Zhang, University at Buffalo
Jee Eun Kang, University at Buffalo
Changhyun Kwon, University of South Florida

Abstract: We develop a multi-period capacitated flow refueling location problem for electric vehicles (EVs) as EV market responds to the charging infrastructure. The optimization model will help us deter mine the optimal location of chargers as well as the number of charging modules at each station over multiple time periods. We define a number of demand dynamics (varying degree of sensitivity to path specific charging opportunities and general charging opportunities) with two objective functions (one maximizing flow coverage and the other maximizing electric vehicle demand). A case 7 study based on a road network around Washington, D.C., New York City, and Boston is presented to provide numerical experiments related to demand dynamics, illustrating the potential outcomes of multi-period charging infrastructure planning.

10:15 AM- 12:00 PM
Lecture Program: Vulnerable Road Users: Analysis of Pedestrian and Bicycle Crashes

Preliminary Investigation of the Effectiveness of High Visibility Crosswalks on Pedestrian Safety Using Crash Surrogates

Md Tawfiq Sarwar, NRC Research Associateship
Grigorios Fountas, University at Buffalo
Courtney Bentley, University at Buffalo
Panagiotis Anastasopoulos, University at Buffalo
Alan Blatt, CUBRC
John Pierowicz, CUBRC
Kevin Majka, CUBRC
Robert Limoges, New York State Department of Transportation

Abstract: Using the SHRP2 naturalistic driving study (NDS) data, this paper provides a preliminary evaluation of the effectiveness of high-visibility crosswalks (HVC) in terms of improving pedestrian safety at uncontrolled locations.  This is accomplished by analyzing the driving behavior of SHRP2 participants at three uncontrolled locations in the Erie County, New York test site.  In this context, crash surrogates (i.e., speed, acceleration, throttle pedal actuation, and brake application) are used, in order to evaluate the participants’ driving behavior, primarily on the basis of data before and after the HVC installation.  The before/after analysis allows the assessment of HVC effectiveness in terms of driver behavior modification.  Mixed logit and random parameters linear regression models are estimated, and panel effects and unobserved heterogeneity are accounted for.  A number of factors are explored and controlled for (e.g., vehicle and driver characteristics, roadside environment, weather conditions, etc.), and the preliminary exploratory results show that HVCs have the potential to improve pedestrian safety and positively modify driving behavior.

1:30 PM- 3:15 PM
Lecture Program: Artificial Intelligence Methods and Modeling Tools for Classification, Estimation,

Interval Prediction of Short-Term Traffic Volume Based on Extreme Learning Machine and Particle Swarm Optimization

Lei Lin, PARC Research
John Handley, Xerox Corporation
Adel Sadek, University at Buffalo

Abstract: Short-term traffic volume prediction models have been extensively studied in the past few decades. However, most of the previous studies only focus on single-value prediction. Considering the uncertain and chaotic nature of the transportation system, an accurate and reliable prediction interval with upper and lower bounds may be better than a single point value for transportation management. In this paper, we introduce a neural network model called Extreme Learning Machine (ELM) for interval prediction of short-term traffic volume and improve it with the heuristic particle swarm optimization algorithm (PSO). The hybrid PSO-ELM model can generate the prediction intervals under different confidence levels and guarantee the quality by minimizing a multi-objective function which considers two criteria reliability and interval sharpness. The PSO-ELM models are built based on an hourly traffic dataset and compared with ARMA and Kalman Filter models. The results show that ARMA models are the worst for all confidence levels, and the PSO-ELM models are comparable with Kalman Filter from the aspects of reliability and narrowness of the intervals, although the parameters of PSO-ELM are fixed once the training is done while Kalman Filter is updated in an on-line approach. Additionally, only the PSO-ELMs are able to produce intervals with coverage probabilities higher than or equal to the confidence levels. For the points outside of the prediction levels given by PSO-ELMs, they lie very close to the bounds.

Wednesday, January 11th, 2017

8:00 AM- 9:45 AM
Lectern Program: Technology Use in Travel Surveys

Inferring Activity-Mobility Behavior of College Students Based on Smartcard Transaction Data

Negin Ebadi, University at Buffalo
Jee Eun Kang, University at Buffalo
Samiul Hasan, University of Central Florida

Abstract: Understanding individual activity-mobility behavior at a finer spatio-temporal resolution has various applications including urban planning, traffic management, spread of biological and mobile viruses, and disaster management. In recent years, proliferation of modern data sources such as GPS observations, mobile phone call records, smart card transactions, and social media activities significantly improved the quality of the activity-mobility pattern observations and reduced the cost of data collection. In this research, we propose to use UB card as a convenient source of combined data in order to define a campus-wide model for constructing students' activity-mobility trajectories in time-space dimension. UB Card is a student's official ID at the University at Buffalo and is used across campus for various activities including Stampedes and Shuttles (on-campus bus system), facilities access, library services, dining and shopping. Therefore, it could be a reliable source of data to identify time, location and activity types of individual students.

In this paper, we present two activity-mobility trajectory reconstruction algorithms. The base algorithm constructs students' activity-mobility patterns in space-time dimension using a set of smart card transaction data points as the only inputs. Then we modified the base algorithm to construct activity-mobility patterns with prior knowledge of students' previous patterns as they have similar patterns for certain days of the week. A database of 37 students' travel survey and UB card transactions that contains a period of 5 days have been used to illustrate the results of the study. These Travel surveys contain detailed information of the students' daily routine from home to school and back as well as other activities such as social, shopping, exercise, etc, that is used to validate the performance of these algorithms.
Three measures of errors have been proposed to capture the time allocation, location deviation, and activity sequences. These errors present an acceptable accuracy (12-25\% error ranges for activity types and average 0.04-0.16 miles of error for location predictions) and show the potential of inferring activity-mobility behaviors based on transaction type data sets.

 

10:15 AM- 12:00 PM

Poster Program: Transportation Demand Forecasting Poster Mega-session, Part 2 (Part 1, Session 829)

A Random Utility Based Estimation Framework for the Household Activity Pattern Problem

Zhiheng Xu, University at Buffalo
Jee Eun Kang, University at Buffalo
Roger Chen, Rochester Institute of Technology (RIT)

Abstract: This paper develops a random utility based estimation framework for the Household Activity Pattern Problem (HAPP). Based on the realization that outputs of complex activity-travel decisions form a continuous pattern in space-time dimension, the estimation framework is treated as a pattern selection problem. In particular, we define a variant of HAPP that has capabilities of forecasting activity selection and durations in addition to activity sequencing. The framework is comprised of three steps, (i) choice set generation, (ii) choice set individualization and (iii) multinomial logit estimation. The estimation results show that utilities for work, shopping and dis-utilities for travel time, time outside home, and average tour delay are found to be significant in activity-travel decision making.