Transporation Operations Laboratory: Article I - Modeling Transportation Systems: Past, Present, and Future
FHWA's new Concepts and Analysis testbed will advance visualization of traffic networks and strategies to help researchers improve safety, mobility, and performance.
Transportation professionals use computer simulations to study various highway concerns and complicated traffic relationships. The models use analytical or numerical procedures to create and evaluate future transportation designs and concepts before they are built. As new transportation challenges arise and new countermeasures are proposed, however, enhancements need to be incorporated into simulation tools to replicate issues accurately and enable researchers to draw valid conclusions.
The Federal Highway Administration's (FHWA) Turner-Fairbank Highway Research Center (TFHRC) initiated research on transportation modeling and simulation almost 40 years ago. In the 1990s, TFHRC researchers developed CORridor SIMulation (CORSIM), a widely used traffic simulation software program that is applicable to surface streets, freeways, and integrated transportation networks with a variety of control devices such as stop and yield signs, traffic signals, and ramp metering. Since then, researchers at TFHRC have been leading and overseeing numerous transportation simulation studies.
"FHWA is committed to continuing as a key contributor to future modeling and simulation research and will lead the development of computer simulation capabilities with the help of TFHRC's new Transportation Operations Laboratory [TOL]," says Dr. Joseph I. Peters, director of FHWA's Office of Operations Research and Development (R&D). One component of the TOL is the new Concepts and Analysis testbed. This testbed will incorporate a repository of transportation models at various levels to allow computer simulations and visualizations of representative traffic networks and experimental strategies to improve safety, mobility, and environmental performance.
But first, a look at the history of modeling and simulation research at TFHRC, plus a review of several current simulation studies.
History of Modeling and Simulation at TFHRC
In the early 1970s, FHWA led the development of NETwork SIMulation (NETSIM), a microscopic traffic simulation model for urban networks, and, in the 1980s, the development of FREeway SIMulation (FRESIM), a microscopic simulation for freeways. Microscopic simulation tracks the movement of individual vehicles, such as car-following and lane-changing behavior, as vehicles move through a traffic network. A microscopic simulation can be used to analyze, for example, key bottlenecks on freeways where the movement of individual vehicles on separate lanes needs to be represented.
Two decades later, researchers at TFHRC, with assistance and participation from universities and private industry, merged NETSIM and FRESIM into a single microscopic model, CORSIM, and developed the Traffic Software Integrated System (TSIS) package, which is a collection of software tools for use by traffic engineers and researchers. At the same time, several universities developed their own microscopic simulation tools for research purposes; however, TSIS/CORSIM was the only viable microscopic traffic simulation model available to practitioners.
By the late 1990s, a number of commercial vendors began offering their own versions of microscopic traffic simulation packages to meet the growing demand. Today, the popularity of microscopic simulation packages continues to increase, and a viable market now exists for commercial traffic simulations.
In the early 2000s, FHWA reevaluated its role in the traffic simulation market. As a result of this assessment, the agency decided to take a different role. Rather than compete with commercial simulation vendors by continuing to develop TSIS/CORSIM, the agency would act in a market facilitator role by focusing public resources on fostering an environment of public-private cooperation through research products that benefit the entire traffic simulation community of practitioners, vendors, and researchers.
During that time, TFHRC also led efforts to develop a variety of traffic analysis models. Major products include the Intelligent Transportation System (ITS) Deployment Analysis System tool, which helps planners analyze the costs and benefits of ITS investments; the QuickZone package for work zone planning and traffic analysis; and the DYnamic Network Assignment-Simulation Model for Advanced Road Telematics (DYNASMART), which supports network planning and operations decisions. Telematics is the integration of wireless communications and positioning systems technology.
In addition, TFHRC has conducted research that produced the first comprehensive and fully validated family of models that facilitated the parallel research, development, and testing of advanced traffic control systems. These models not only led to the development of traffic control systems but also enabled practicing traffic engineers to conduct operations analyses.
Currently, researchers at TFHRC continue to lead studies that use various types of simulation tools. For example, four recent research projects use computer simulations to examine topics related to driver behavior, work zones, roundabouts, and autonomous intersections.
Driver Behavior in Traffic
Existing traffic analysis and simulation tools cannot effectively model drivers' abilities to recognize and respond to their environment with behavior appropriate to the encountered driving situation. This research project, which began in 2009 and is funded under FHWA's Exploratory Advanced Research Program, aims to answer a number of questions related to driver and traffic performance: driving rules during normal and abnormal driving conditions, the magnitude of difference in driving rules practiced by different motorists, the impact of an incident on system performance, and the effect of drivers' interactions during incidents.
This study aims to characterize driver behavior using a naturalistic driving database (data captured from volunteer drivers' daily driving) and agent-based modeling techniques, which simulate the actions and interactions of autonomous agents such as drivers to assess their effects on the system as a whole. The project is developing intelligent agents (software representations of drivers) that are designed to learn drivers' temporal decisions in response to varying traffic situations retrieved from the naturalistic driving database. The driving rules of the agents will be coded in a computer simulation tool to test and study the collective effects of learned behaviors with multiple drivers under different situations.
The study uses reinforcement learning, a novel and successful area of artificial intelligence, to tackle how an independent agent that senses and acts on its environment can learn to choose logical actions to reach its long-term goals. This method enables the agent to keep learning from observations, actions conducted, and rewards received. At the conclusion of this project, expected at the end of 2011, agents will be developed to mimic realistic driver behaviors in various traffic scenarios. After verification and validation of the developed agents, the FHWA researchers will embed an abstraction of the agents' learned driving rules in VISSIM®, a microscopic traffic simulation tool. An example of a learned driving rule is the gap maintained between the agent's vehicle and the vehicle in front. Another example is the time taken to decelerate a vehicle when approaching a congested area or a red light.
Travel Patterns and Work Zones
Highway construction has become a main source of congestion, and traffic simulation models are widely used to analyze traffic problems related to work zones and to evaluate mitigation strategies. Traffic simulation models, especially microscopic ones, have become popular because they provide a controlled environment in which researchers can analyze and evaluate a wide range of scenarios, including highway bottlenecks, complex geometric configurations, and operational improvements.
Most work zone studies tend to focus on analyzing local traffic impacts. Small-scale work zones may be analyzed successfully without the computationally expensive iterations needed to reach an equilibrium condition (a balance of "origin trips" and "destination trips" in the simulated traffic network). But long-term and large-scale studies require analyses with an equilibrium approach.
Unlike other microscopic simulation models, this project used the TRansportation ANalysis SIMulation System (TRANSIMS) produced by FHWA's Travel Model Improvement Program to analyze changes in travel patterns and mobility impacts caused by work zones. As TRANSIMS is capable of analyzing the dynamic nature of a traffic stream as well as travelers' activity patterns, it could be a viable option for analyzing the impacts of major work zones. Especially in the case of long-term highway closures, understanding not only mobility impacts but also travelers' day-to-day route changes is important.
For a case study, the researchers built a TRANSIMS model for the southeast Michigan area, including Detroit and the surrounding seven counties, with a population of 4.9 million in 2000. By applying a day-by-day evolutionary approach, the TRANSIMS model investigated mobility impacts, travel pattern changes, and departure time shifts with each freeway segment closure.
The highway work zone under investigation was the I-75/I-96 Ambassador Gateway Bridge reconstruction. During the reconstruction, the I-75 mainline and I-75/I-96 system interchange were closed for months. The scenario included complete closure of I-75 near the Ambassador Bridge area and the I-75/I-96 system interchange.
Using the TRANSIMS model, the researchers were able to analyze travel pattern changes as well as travel reliability. They also investigated departure time shifts as an impact of highway work zones. With a simple choice model for departure times, the researchers analyzed shifts by type of traveler.
"This project demonstrated that TRANSIMS is a viable tool for traffic analysis of work zones," says Brian Gardner, leader of the transportation systems performance team in FHWA's Office of Planning and manager of FHWA's TRANSIMS program. "The study's approach sheds light on behavioral analysis as a component of work zone impacts."
Calibration of Two-Lane Roundabouts
Accurate and realistic traffic simulation can help improve the reliability of analyses related to isolated roundabouts. Furthermore, it can be more reliable in analyzing road networks that combine signalized and unsignalized intersections, including interactions from spillback (traffic queuing at an intersection) and headways (time difference between consecutive vehicles).
The FHWA researchers studied three existing two-lane roundabouts in Malta, NY, to obtain data for reliable calibration of the VISSIM software. The researchers differentiated the data by interior and exterior lanes for the roundabouts' entry and circulatory lanes. They also studied multiple parameters in VISSIM to obtain more realistic results, including temporal gaps between vehicles, speed profiles (approaching and circulating speeds), and model parameters for following cars.
Results from those three factors (but mainly time gap) and priority rules in VISSIM show similarities with results reported in the National Cooperative Highway Research Program's NCHRP 572: Roundabouts in the United States. The capacity model described in NCHRP 572 was developed from U.S. field data from four sites in Maryland, Vermont, and Washington at seven approaches during congested or queuing periods only. The NCHRP 572 models used exponential functions to best fit field-collected capacity data.
The capacities from the NCHRP models based on field data are comparable to the VISSIM calibrated results along most of the circulating volumes, except at the upper boundary. The diverging differences after circulating volumes of about 1,400 vehicles per hour (vph) are due to a lack of field data points collected or available for higher circulating flows. This likely bias is caused by the model form being extended beyond available data. A user of the NCHRP model is likely to overestimate capacity beyond 1,400 vph, which could lead to congestion prior to the end of the life cycle of the proposed roundabout. In brief, calibrated traffic simulation for roundabouts seems to be more reliable than deterministic models under most traffic volume conditions. Therefore, calibrated simulations also are more reliable for network traffic analysis.
Autonomous Control At Intersections
As population growth has gone beyond transportation systems' abilities to handle increased levels of demand, congestion has become one of the most challenging engineering issues today. The roadway system is a source of mobility not only for drivers, but also for goods and services. As such, the system's ability to handle vehicle demand is paramount to the Nation's economic prosperity. Traffic congestion accounted for an estimated $115 billion in losses in 2009. Facing limited budgets and available right-of-way to build excess capacity, transportation professionals are looking to operational improvements to address congestion.
Human error accounts for 70 to 80 percent of vehicle crashes. Prevention of those crashes -- and reduction of traffic congestion -- through semi-automated, and ultimately automated, driving is achievable, but requires new systems to coordinate the movement of autonomous vehicles in complex traffic situations. Recent research in artificial intelligence and robotics continues to make the feasibility of automated vehicles a much more tangible reality than it was in the past. Although not at a stage warranting mass deployment at present, this technology clearly has the potential to eventually become the standard.
Examining the consequences of what could amount to a major overhaul of traditional operating systems is therefore critical before automation is introduced. Just as congestion mitigation is an important societal problem, the operational efficiency derived from implementations aimed at exploiting the technological advantages of automated vehicles warrants serious consideration.
Toward that end, another current project under the Exploratory Advanced Research Program is examining the feasibility of autonomous vehicles and intersections for use within the next 20 years. This project is examining a new form of intersection control that can increase vehicle throughput dramatically by taking advantage of the capacity of autonomous vehicles. The study uses an automated mechanism for intersection control using a new first-come-first-served protocol that has the potential to process traffic much more efficiently than traffic signals -- without compromising safety. Its development is guided by criteria that include the use of sensor technologies, adoption of a standardized communication protocol, and the ability to deploy incrementally, allowing expansion to other intersections and adaptation to increasing numbers of autonomous vehicles. Absolute collision prevention, even under conditions of communications failure, is the primary goal.
Improved intersection management will be a major step toward an infrastructure for fully autonomous vehicles that will revolutionize transportation of people and goods. To test the performance of protocols for automated intersection control, the use of microscopic simulation models becomes indispensible. Because the technology for autonomous vehicles currently is not at the level needed for real-world testing with any meaningful traffic flow, microscopic simulation is necessary to obtain an estimate of the performance of systems of autonomous vehicles. The FHWA project team developed and implemented a microscopic simulation tool from scratch so researchers can model the intersection control system and evaluate its performance.
The project team conducted experiments using a population of autonomous vehicles to compare the performance of an intersection outfitted with the automated first-come-first-served protocol versus an intersection with a traditional traffic signal. The results show that the first-come-first-served strategy performs significantly better than a traditional traffic signal, reducing average vehicle delay by an order of magnitude in all cases. The results are encouraging, but FHWA plans to conduct additional research to further validate the first-come-first-served intersection control strategy and to develop more efficient intersection control systems.
Transportation Operations Laboratory
Followup studies to these research projects are planned to be carried out in the Transportation Operations Laboratory at TFHRC. The lab will consist of three components: (1) a Concepts and Analysis testbed, (2) a Data Resources testbed, and (3) a Cooperative Vehicle-Highway testbed.
The testbeds are intended to provide FHWA researchers and others in the transportation commu-nity with innovative, reliable, and accessible pools of resources to conduct research and tests, create visualizations, and present quality research findings in a dependable and cost-effective manner. Future articles in Public Roads will describe the Data Resources and Cooperative-Vehicle Highway testbeds.
Concepts and Analysis Testbed
The Concepts and Analysis testbed will incorporate a repository of transportation models to allow computer simulation runs at the three levels of analysis -- micro-, meso-, and macroscopic. As mentioned earlier, microscopic models track the movement of individual vehicles, such as car-following, as vehicles move through a network. Macroscopic simulations, on the other hand, model large geographic areas to simulate traffic flows, speeds, and densities. Mesoscopic models, such as DYNASMART, represent an intermediate level of analysis between micro- and macroscopic models, enabling simulation of individual vehicles at a corridor or regional level with visualizations that can reveal greater insights about congestion dynamics and sources of congestion. The visualizations will improve model performance. The outputs of simulation analyses can, for example, provide an authoritative basis for conducting benefit-cost analyses of experimental strategies.
Simulation Tool for Autonomous Control Project
The Concepts and Analysis test-bed will support and benefit research being conducted by FHWA's emerging ITS and Exploratory Advanced Research projects, plus external needs from academia and other research institutes. Expectations for this testbed and the Transportation Operations Laboratory include the following:
- Enable FHWA research staff to refine experimental strategies through direct interaction with the models and to determine alternative strategies' potential value to various stakeholders.
- Provide a repository of authoritative and representative model application datasets that will enable researchers to compare technical and performance measures in realistic scenarios and across projects, regardless of partners and perhaps even choice of model platform, so that valid assessments can be made.
- Provide a venue for hosting technical and outreach meetings at which the above results could be reviewed and discussed by producers, FHWA research staff, other U.S. Department of Transportation staff, and stakeholders.
- Provide resources that consultants supporting FHWA research projects can use to improve the consistency, quality, efficiency, transparency, and availability of research results and government deliverables.
- Afford a venue and resources to support ongoing project work by FHWA's Traffic Analysis Tools team, the Traffic Analysis and Simulation Pooled Fund Study, and other internal research initiatives.
- Provide a venue and resources to support research by students and other outside research staff who can use the models for their own projects, with benefits to the larger community.
The Testbed's Potential Applications
The Concepts and Analysis testbed provides a needed resource for many of FHWA's new ITS programs, such as Dynamic Mobility Applications and Applications for the Environment: Real-Time Information Synthesis programs. Other projects that stand to benefit include those under FHWA's Exploratory Advanced Research Program and traffic simulation research such as for innovative highway designs.
Dynamic Mobility Application Program. This program seeks to identify, develop, and deploy applications that leverage the full potential of connected vehicles, travelers, and infrastructure to enhance current operational practices and transform future management of surface transportation systems. Under proof-of-concept testing and tool development, the program will begin to assess innovative applications with potential to improve transportation operations. The FHWA researchers will test promising applications using simulated testbeds, which will assist in the assessment of whether specific applications can be expected to perform well in early deployment stages.
Applications for the Environment: Real-Time Information Synthesis Program. The overall objective of this program is to generate and acquire environmentally relevant real-time transportation data and innovative applications, and use them to create actionable information to support and facilitate green transportation choices by users and operators. Employing a multimodal approach, the program will work in partnership with other research efforts to better define how data and applications might contribute to mitigating some of the negative environmental impacts of surface transportation. The researchers will use sophisticated models and other analysis tools to determine how the proposed environmental improvement strategies will work, their effectiveness, how they compare to each other, and how they compare with strategies developed by other researchers to improve mobility and safety. The program will analyze potential evaluation tools, build an evaluation process, and develop baseline estimates for evaluating environmental assessment and improvement strategies.
Exploratory Advanced Research Program. Several of the current projects sponsored by this program are using simulation models to support ongoing work or else are developing and assessing innovative concepts that might benefit from follow-on research where the concepts can be simulated.
Innovative Intersections and Interchanges. Through in-house research, FHWA has explored, studied, and marketed a number of innovative designs for intersections and interchanges. As a result, several States are constructing many successful designs such as the double crossover diamond interchange and the displaced left-turn intersection. FHWA will study additional innovative designs and applications along corridors using traffic simulation to evaluate performance and visualize geometric layouts and interactions of drivers and vehicles.
As new transportation challenges arise and countermeasures are proposed, computer simulation plays a critical role by enabling researchers and practitioners to evaluate innovative designs and ideas. When the Transportation Operations Laboratory comes fully online in the near future, it will have the capability to host a variety of modeling and simulation studies so researchers can use a full range of analysis tools to assess the viability of potential solutions to tomorrow's transportation challenges.
Joe Bared, Ph.D., P.E., is team leader for the Transportation Operations Concepts and Analysis Team in FHWA's Office of Operations R&D. He has worked at FHWA for more than 20 years and managed the program area on intersection and interchange safety and operational effects of design. He managed development of the first roundabout guide in the United States and has promoted innovative intersection and interchange designs in a new FHWA publication, Alternative Intersections/Interchanges: Informational Report (FHWA-HRT-09-060).
C. Y. David Yang, Ph.D., is a research engineer with FHWA's Office of Operations R&D. He is responsible for traffic modeling and simulation research at TFHRC. He is chairman of the Transportation Research Board's User Information Systems Committee and serves on the editorial board of the Journal of Intelligent Transportation Systems: Technology, Planning, and Operations. He received his bachelor's, master's, and doctoral degrees in civil engineering from Purdue University.
Peter Huang, Ph.D., P.E., is a research engineer with the FHWA Office of Operations R&D. He has more than 25 years of experience in ITS and has led many ITS-related research and deployment projects. Previously, he was an associate professor with Beijing Jiaotong University in China, a senior ITS engineer with TRW Corp. and EG&G Corp., and ITS engineering manager with the Utah Department of Transportation. He has a doctoral degree in transportation engineering from the University of Cincinnati.
Randall S. VanGorder is a research engineer with FHWA's Office of Operations R&D. He has more than 20 years' experience in transportation and manages the Traffic Analysis and Simulation Pooled Fund Study, comanages the Traffic Management Center Pooled Fund Study and manages the day-to-day operations of the Traffic Research Laboratory. He received a bachelor's degree in civil engineering from Penn State University.
The authors would like to thank Robert Ferlis, technical director of FHWA's Office of Operations R&D for his encouragement during the preparation of this article and his suggestions on the content.
For more information, contact C. Y. David Yang at 202-493-3284 or firstname.lastname@example.org.