CMV Safety | Freight Logistics & Policy | Truck Size & Weight | TACT | Wireless, Vehicle-to-Infrastructure Communication

A PRELIMINARY CONCEPTUAL APPROACH TO THE TRB RESEARCH NEED STATEMENT ON DATA VISUALIZATION AS APPLIED TO FREIGHT OPERATIONS

Sowmya Karthikeyan
ALK Technologies, Inc.
1000 Herrontown Road
Princeton, NJ 08540
609 252 8198
skarthikeyan@alk.com

Ronald G. Hughes, Ph.D.(1)
Program Director
Visual Analytics, Modeling and Simulation Research Group
Institute for Transportation Research and Education
North Carolina State University
Centennial Campus, Box 8601
Raleigh, NC 27695-8601
919 515 8523
ron_hughes@ncsu.edu


Dr. Alain Kornhauser
Professor of Operations Research & Financial Engineering
Director, Transportation Program
Faculty Advisor, PAVE (Princeton Autonomous Vehicle Engineering)
Department of Operations Research & Financial Engineering
229 Sherrerd Hall (ORFE Building).
Phone: 609-258-4657
alaink@princeton.edu

Submitted July 26, 2010

Word Count: 4536 with 7 Figures

(1)Corresponding Author

ABSTRACT

The present paper puts forth a conceptual framework for the application of visualization to freight data. This is in response to a Transportation Research Board (TRB) Research Need Statement: Application of Data Visualization and Visual Analytics to Freight Operations and Logistics at the Sub-National Level. Reference is made to a similar application in the health communications field and to the use of XML (extensible markup language) as the possible platform requirement for such an application. While specific "displays. are not recommended at this time, the general form and function of a graphic user interface is suggested. An effective application of data visualization in this area rests upon the identification of the specific user populations involved and their unique operational needs. The paper concludes with reference to a current state level situation where public opposition to the development of increased port capacity "trumps. analysis data showing significant economic benefit . . . indicating that traditional applications of visualization in the public involvement process are still required to mediate public perceptions of "freight. and analytical projections of its social and economic benefits.

INTRODUCTION

The TRB standing committee on Visualization in Transportation (ABJ95) is one of eighteen committees in the Data and information Technology Section. The scope of the Visualization Committee is to foster and disseminate collaborative exchange and research that enhances the usable knowledge of visualization methods and technologies for their potential in addressing critical transportation issues of today, as well as promoting innovative approaches to society's transportation needs of the future.

Visualization is increasingly recognized as a "cross cutting. discipline within the transportation field (1). Within TRB, the technical committees generate research need statements (2), the aim of which is to stimulate research that addresses concerns, issues, or problems facing the transportation community. Research needs identified by the Visualization Committee address both "applications. of visualization as well as more basic concerns related to the further development of visualization as a tool.

Eight of the fourteen current TRB Research Need Statements in visualization are from ABJ95; the remainder are from other TRB standing committees: AHB70 (Access Management), ABJ60 (Geographic Information Science and Applications), ABJ20 (Statewide Transportation Data and Information Systems), AFH30 (Emerging Technology for Design and Construction), and AFB40 (Landscape and Environmental Design). The full text of each statement can be viewed online at:

http://rns.trb.org/advanced_project.asp?f1=k%3A%3AKeywords+%28Title+or+Description%29&ddlType=RNS&orgType=S&status=&date_params=&lower_date=1900&upper_date=2099&sb=&so=a%3A%3AAscending&sc=xx%3A%3AAll+Categories&t1=visualization

There have been two NCHRP "state of the practice. reviews, or "synthesis studies,. of visualization since 1996 The first, published by Landphair and Larsen (3) focused on the emerging application of advanced computer image generation and computer graphics for generating photo-realistic images for use primarily in the public involvement process. Ten years later, Hixon (4) conducted a second "synthesis. of the current state of practice where he indicated that, despite the continued absence of definitive guidance for practitioners and project management personnel, and despite the continued lack of benefit-cost data addressing the cost effectiveness of visualization, there was a growing use of visualization within state departments of transportation and the supporting engineering consulting community.

In 2007, the TRB publication TR News, devoted a special issue to visualization in Transportation (1). In 2008, three members of ABJ95 (Rhyne, Hughes, and Manore) conducted a "class. on visualization in transportation at the IEEE ACM SIGGRAPH International Conference in Los Angeles, CA, further indicating the cross cutting aspects of visualization applications and the role of computer science in future efforts (5).

Visualization interest within TRB has grown from a sub-committee, to a task force, to the status of a standing committee. TRB has hosted international symposia on visualization in Minneapolis, MN, Orlando, FL, Houston, TX, and Salt Lake City, UT. In 2007, concurrent with achieving the status of a standing committee with TRB, the visualization committee published its first set of "research need statements..(6) It is the opinion of the ABJ95 Research Needs Sub Committee that these statements remain "valid.. We still require usable "guidance. for visualization from "users. and project managers; we still require a clear benefit-cost template by which to evaluate the effectiveness of visualization across various applications; we continue to need information and guidance on how to integrate the different sources of spatially referenced data; we still need an education and training "curriculum. for preparing users and practitioners; we still need more guidance on "planning. oriented applications of visualization; and we need research emphasis on "data. visualization (for demand modeling, for freight and logistics, etc.); and in general more emphasis on the need to represent the underlying "dynamics. of system operation (i.e., via modeling and simulation approaches).

An important use of visualizing data is to understand and probe it by finding patterns in it to ask the right questions.(7) The typical human brain is more adept at processing new information when presented visually, and is more readily able to compare and contrast this new information with its accumulated knowledge base. Where the new information correlates with the history of knowledge, it leads to confidence in the validity of the underlying data. If the new information is different from what the brain expects, within the context of historic knowledge, it can lead to asking the right questions for deriving additional new information from the data (colloquially referred to as "drilling down the data") and either learning from it, i.e. changing one.s perspective and adding to the knowledge base, or identifying problems to be resolved in the data.

Understanding freight transportation data is vital for evaluating major investments and their implications on the flow of goods.(8) Whether a potential investment has a differential effect geographically or economically, or has the same impact on everyone is central to policymaking. Since transportation data has an inherent geographic and spatial aspect to it, as does accumulated contextual knowledge, it is more fitting for visual data analysis.

FOCUS OF PRESENT PAPER

The present paper focuses on a visualization research need submitted in 2009 by ABJ95 at the urging of the freight data subcommittee and by those working on SHRP 2 in the area of freight and logistics. Freight and logistics represent two potential "application areas. for data visualization. They also provide the opportunity to address one of ABJ95.s original research needs, that being to "visualize. through modeling, simulation, and data visualization the operational dynamics of system performance.

Visualization work in the area of freight and logistics are in their infancy.(9, 10) Current efforts (11, 12) focus in large part of extending the application of fleet GPS data for mapping major truck corridors and commodity flow. The purpose of the present paper is to share the thinking of those attempting to respond to this need from the standpoint of establishing some sort of more "structured thinking. about the domain of freight and logistics and what it might mean to "visualize. both the structural and dynamic/operational aspects of these important transportation system concepts (13). Current "thinking. encompasses not only the need to identify key data elements/performance indicators, the total range of user needs, but also the system architecture requirements necessary to handle extremely large, diverse data sets. One thing we can be fairly certain about is that the end product of this work will be well beyond the "Excel-to-Chart. type of solution to which we are presently accustomed.

Figure 1 Solutions to the Freight Data Visualization Problem are not Likely to Resemble the Current "Excel-to-Chart. Type Solution

BACKGROUND

First, let's establish what we mean by a freight 'system.' Figure 2 provides a schematic view of a typical freight transportation model. Omitted from this particular schematic view is the 'aviation' component, which if included would be characterized by the same types of relationships depicted here for rail, road, and water networks.

Figure 2 A Simplified Schematic and Explanation of a Generic Freight Transportation System Model

Such a schematic view (or visualization) is useful for orienting individuals to the basic structural and functions aspects of what we mean by a freight 'system.' It defines the 'scope' of the problem in terms of the types of data required to represent system performance; basically the modal demands, the role of transfer points, system outputs by modes, and the notion of 'transfers' that take place within the system. It suggests the need to understand and to quantify the 'impedances' on input, transfer, and output operations.

Figure 3 provides a somewhat more 'pictorial' view of the 'infrastructure' elements of a hypothetical statewide freight system. The figure attempts to illustrate freight entering the state through seaport (SP) facilities; the presence of 'short sea' shipping operations with other port facilities; the presence of port-side rail connections and the suggestion of intermodal and double stacked freight rail to other sea as well as inland facilities; the notion of major and minor industrial 'feeder' zones/clusters and related work force feeder zones; the presence of air freight facilities, etc. Like Figure 2, the visualization provides information about the overall 'structure' of the freight system, but in and of itself provides no ability to interact with the visualization in terms of its operational performance.

Figure 3 Visualizing the Major Infrastructure Components of a Statewide Freight System

The suggestion is that by "clicking. on a facility, one could introduce some sort of pull down or pop up window that would allow the user to obtain a historic or current snapshot of that facility.s performance (e.g., commodities, by weight, value, time of year, etc.). Such snapshots may or may not provide information on important intermodal attributes (i.e., the transfer of goods from one mode to another). The visualization of a "terminal. per se conveys nothing about the nature of the transfers, the times or costs associated with the transfers, etc. Such visualizations provide "visual archives. of what we know about the structural elements of the system, but do not permit the user/viewer to interactively manipulate the functional attributes of the system; i.e., the probe the important "what if. questions of system performance.

Figure 4 is presented here for the benefit of the visualization community who, at least within the transportation environment, has till now focused almost exclusively on the structural elements of the system, and almost exclusively on those transportation system components dedicated to the movement of people. Figure 4 breaks down transportation systems into those systems whose primary objective is to move "people. and those whose primary objective is to move "goods.. The figure identifies elements that are common to both types of system, and further into those elements that are "structural. and those that are "dynamic..

Until recently, the domain of transportation visualization efforts has been primarily defined by a focus on the visual representation of the "structural. elements of the system (the form, if you will) as opposed to the "function. (i.e., how the system operates). The TRB Visualization Committee (ABJ95) in its TRB Research Need Statements has recognized the need to place increased emphasis on the use of modeling, simulation, and in some cases "4D. visualization to represent the operation of data intensive systems. Focusing on the dynamic, operational aspects of system performance fall more into the area of "data visualization.. (11). It is our belief that the visualization requirement for freight and logistics is more aligned with data visualization than with the traditional 3D visual representation of system components.

We are increasingly recognizing that it is desirable or even essential, that functional simulations or models of freight system operation have an "economic. component. For the legislator seeking to use the model or simulation to determine the impacts of investment on the ability to attract new industry to area, to create new jobs, or to generate increased state revenues, an economic component is essential. The ability to address such "standard. economic questions should be embedded in the model or simulation, as well as a simple "visual. means of evaluating such impacts.

Figure 4 Transportation Systems to Move People and Those to Move Goods: Their Common System Variables, Their Structural Components and Their Dynamic Components

In the material that follows, we take a preliminary look at what data visualization of a freight system might look like in terms of the types of data required, the key functional relationships that need to be captured- including the economic- and a simple but effective user visual interface (the data visualization component, if you will) between the data and the "user..

An 'Example' from the Health Communications Field and its Relevance to the Visualization of Freight Data

As an example, we point to a data visualization approach by GE HealthyMagination reported on line at: (http://www.ge.com/visualization/your_health_visualizer/). In the example shown below, the intent is to communicate to the general public the relationship between one.s own health "profile. and the likelihood of major health problems. In both the top and bottom examples, we have selected 'male' and 'non smoker.' The application permits the user to use the sliding Body Mass Index scales to individually set height and weight. In the above example, 5ft 11 inches and 179 pounds is shown as a 'normal BMI. At the top are two bars; the darker colored bar shows the percent of individuals who, with the same or similar profile, have been diagnosed with the disease in question (Type II Diabetes, Hypertension, Heart Disease, or Stroke). The lighter colored bar shows the percent of individuals with different height/weight profiles who are diagnosed with the disease.

By adjusting (sliding) ones height and weight attributes left or right, one is able to easily determine what combinations of height and weight place him in either the 'normal BMI' range or the 'obese' range and how these changes affect their predicted likelihood of being diagnosed with the health condition in question. The user can also explore the data with regard to the impact of being a 'smoker' or an 'ex smoker.' The application might be improved for those wanting to see a more continuous, graphical representation of these results by allowing the viewer to capture the percentage values associated with different inputs and by allowing the user to graphical represent/plot these values in order to see a more "continuous. view of the functional relationship between these variables.

We are impressed with the simplicity and 'user friendliness' of the application and think that perhaps a similar approach might be effective for visually exploring the functional relationships between key variables in a freight system application. The following represents a preliminary attempt to demonstrate such a conceptual approach.

Figure 5 An Example of Data Visualization in the Health Communications Field from GE HealthyMagination Project

(http://www.ge.com/visualization/ your health visualizer/ )

What is not apparent to the reader in this example is the underlying structure of the data that support the application. In the next section we explore the role of XML in creating an effective platform to enable these types of applications. Keep in mind that the above application, while appearing very "simple. on the surface is generating "the answer. based upon computations being applied to a data set of over 6 million individual profiles based on the unique "patterns. of patient user gender, age, height, and weight.

An Analogous Freight Data Visualization Application

The authors propose that the freight data community works towards creating a similar interactive application to visualize freight data. Figure 6 shows a sample layout of such an application, where the user can probe into the underlying freight data to support their decision making.

Figure 6 Sample Layout for Freight Data Interactive Visualization Application

The interactive application will expose key shipment attributes for the user to query such as mode, commodity, origin, destination and year. This is only an illustrative example, and if such an application were to be developed, additional research will have to be conducted to identify specific data elements from the data source to expose, in order to sufficiently equip the target user to understand the data,without being overwhelmed by the volume of the data elements and intricacies of their relationships.

While popular methods for obtaining user inputs, such as drop down menus and slider scales, can be used, it may be worthwhile to study the recent advancements in computer graphics and other interactive software technologies to incorporate the latest innovative and user-friendly methods to track inputs.

Spatial representation of the data elements is central to understanding transportation data, and is one of the tools highlighted in the visualization application. Although traffic density maps have been used successfully for many years to convey information about the flow of goods, it has three major drawbacks in its current usage. One is the static nature of the output, i.e. it doesn.t allow for the user to quickly compare the displayed flows with different flows to analyze patterns beyond initial observation. Secondly, although it shows volumes over particular network segments very well, it doesn.t relay the individual shipment characteristics of the traffic moving over a segment, such as origins and destinations. For example, what is the mix of traffic over a segment on I-95 in New Jersey that is going to Florida or to North Carolina or coming from a port in New York or from the border in Maine? The third problem is that it is difficult to convey easily the temporal distribution of flows.

The proposed concept for the freight data visualization application addresses all of the above shortcomings. The interactive aspect of the application will resolve the first issue, as it will allow the users to easily change the input settings and view the updated map almost instantaneously. In addition to being able to view selected traffic as it moves over the network, (for example, all traffic originating in a New York, or all interstate truck traffic from all origins to all destinations, moving through North Carolina, or just the intra- state movements in Texas), the proposed tool will allow the user to view individual origin/destinations information of the displayed traffic. One suggestion is to display in a different color, the traffic from all origins to a particular point in the map selected by the user with a simple mouse click. This will not display additional new traffic, but rather display the origin composition of the volumes already displayed on the map. Similarly, a different mouse click could display traffic originating at the user selection to all destinations. A third mouse click option could show all traffic moving via a particular point. For example, if one clicks over a thick bandwidth segment over I-40 in Texas, one can see the geographic extent of traffic moving over that segment. To address the third drawback of viewing temporal changes in traffic, the authors propose using a slider scale for the time element to view changes in traffic over time interactively, or allowing the users to select two time periods and viewing change in flows (for example, increases in red and decreases in green, shown in bandwidths drawn to scale), or if sufficient data is available, allow the user to double-click on a network segment to pop up a temporal distribution chart.

The conceptual interactive tool also includes a selection of simple graphs or charts for the user to view performance indicators or other aggregate information such as tonnage, value of goods, length of haul, on-time delivery, etc. for the selected traffic that is displayed on the map. Dashboards and widgets, commonly used in other industries, could provide additional information in an easy-to-understand format.

Such an interactive tool to visualize currently available data is not very difficult to achieve with the recent advancements in areas of GIS, computer graphics, web-based applications and data transfer technologies, computer processing performance, data storage and access, and cheap memory.

However, prior to developing such tools, it is important to identify the different data sources of freight transportation data that need to be visualized, and also to study the various decision making aspects that this tool will support. This will impact the design of the user selection of data attributes module, and of the outputs to display interactively. Another critical aspect is the design of the data structure modifications and storage to allow for fast querying and computations necessary for creating an interactive user- experience.

One suggestion to enhance development of such tools is the adoption of web-based XML (Extensible Markup Language) format for freight data transfer. Although using an XML-based data format standard is not essential to creating an interactive freight data visualization tool, it will be beneficial in future as more data providers and more software developers work together to create different visualization tools to meet different needs.

For example, consider the following table of data: (stored in any database format - MS Excel, MS Access, SQL, Text, or any proprietary format)

With easily available XML conversion routines, this data can be represented in XML format as:

This data can be queried from another application with a simple command such as:

https://data.freight.gov/tons?mode=rail&commodity=chemicals&origst=texas to get details for all chemicals from Texas moving over Rail.

The advantage of using this format is that the data structure of the individual data sources do not matter, and can be changed to include additional elements without affecting the current users of the data. Additionally, multiple data sources can be easily brought together for a specific application in this format.

If the freight data and modeling community can create a standard XML schema for representing freight data (the tags <origst>, <commodity>, etc), say Uniform Freight Data Language (UFDL) and if data providers adopt this UFDL to transfer freight data, it will advance the pace of development of visualization tools that can use multiple data sources and evolve quickly as data sources grow. A quick web-search on "XML Markup Language" will point to a number of varied industries that currently use this technology.

Figure 7 shows an example overview of freight data transfer between various data sources, models and visualization tools. Geographic, physical and service data for mode-based infrastructure elements, operational data, and data about economic activity, coming from different sources, are used by freight models as well as by visualization tools, using an API (Auxiliary Programming Interface). The outputs of the freight model can be queried using the API by the visualization tool and processed further for creating interactive displays.

Figure 7 Sample Freight Data Transfer Framework

CONCLUDING REMARKS

We began this paper by pointing out the need to address the potential for applying data visualization methods to freight data, suggesting that traditional visualization methods focusing on the visual representation of the more "structural. aspects of a system needed to evolve into applications that focused more on the functional, or operational, aspects of system performance. To that end, we have illustrated examples of such first by reference to data visualization in the GE Healthymagination program and then by a similar conceptual approach to the visualization of freight data. Those expecting to "see. traditional "visualizations. of freight system performance and operations may have been disappointed by the lack of specific examples. Specific examples (visual displays) await a more detailed understanding of the needs/requirements of those who use freight data to draw inferences, make decisions, etc. We feel we have made a major contribution however in pointing out the need to first focus on system level approaches for handling the large and diverse nature of data sets required to deal with the domain of freight, recommending an extensible markup language (XML) approach. There is a computer science component to data visualization applications for freight.

As these concluding remarks are being made, the Raleigh (NC) "News and Observer" (14) is reporting (July 22, 2010) that the N.C. Ports Authority is scrapping plans for a controversial $3 billion international shipping terminal in response to public concern about "potential environmental damage and economic perils" . . . despite claims that the new facility would create 16,000 jobs in loading, trucking, and railroads, as well as from a network of distribution centers built to supply major retail chains. According to the N&O, the proposed port would, according to its opponents, "overwhelm small, tourism- dependent communities with heavy truck traffic, industrial noise, and pollution." In other words, the human level concerns of a laid-back, coastal, retirement and tourist- dependent community "trump. the need for the state as a whole to significantly increase its port capacity in response to the anticipated demand of Post-Panamax container vessels seeking ports on the East Coast of the US. "Freight. can be ugly and intimidating, or at least perceived to be so, when it takes place in one.s own backyard.

While data visualization may permit decision makers to make better informed decisions, there remains a need for traditional visualization methods in mediating the need and public benefit of such facilities with the real and/or perceived personal impact of their development. While data visualization may be necessary to enable us to better understand the complexities of dynamic freight system requirements, it may not be sufficient to assure the approval of those who must consent to its growth and expansion. Traditional visualization methods can help to put a "face. on change. Data visualization can help provide the confidence required to ensure that our predictions about the benefits of such change are firmly anchored to data.

REFERENCES

1. Visualization in Transportation: Empowering Innovation. TR News, September- October 2007, Number 252, Transportation Research Board, Washington, DC.URL for TRB Research Need Statements

2. Landphair, H. and Larson. T. Visualization in Highway Development: Applications of 3D and 4D Visualization Technology in Transportation, NCHRP Synthesis of Highway Practice 229, Transportation Research Board, 1966.

3. Hixon, C. Visualization for Project Development, NCHRP Synthesis 361. Transportation Research Board, Washington, DC. June 2006.

4. The 35th International Conference and Exhibition on Computer Graphics and Interactive Techniques. ACM SIGGRAPH, Los Angeles, CA, August 2008.

5. Hughes, R. Research Agenda of the TRB Visualization in Transportation Committee. In Proceedings of the TRB 5th International Visualization in Transportation Symposium and Workshop, Denver, Colorado, October 2006.

6. Few, Stephen. Visual Pattern Recognition: Meaningful Patters in Quantitative Business Information. COGNOS Innovation Center for Performance Measurement, November 25, 2006 24

7. Banister, D. and Berechman, Y. Transport investment and the promotion of economic growth. Journal of Transport Geography, Volume 9, Issue 3, September 2001, 209-218.

8. Visualization in Transportation: Empowering Innovation. TR News, September- October 2007, Number 252, Transportation Research Board, Washington, DC.

9. Schmitt, R. Dronzek, R., and Rohter, L. Visualization to Solve Problems in Freight Transportation. In Visualization in Transportation: Empowering Innovation. TRNews, September-October 2007, pp 15-19, Transportation Research Board, Washington, DC.

10. TRB Visualization in Transportation Committee. National Academies of Science, Transportation Research Board, Washington, DC. (committee website at: http://www.trbvis.org/MAIN/TRBVIS_HOME.html last accessed on July 24, 2010).

11. Figliozzi, M. and Tufte, K. Prototype for Freight Data Integration and Visualization Using Online Mapping Software: Issues, Applications, and Implications for Data Collection Procedures. TRB 88th Annual Meeting Compendium of Papers DVD, Transportation Research Board, Washington, DC, January 2009.

12. Jones, C, Murray, D., and Short J. Methods of Travel Time Measurement in Freight- Significant Corridors. Submitted to TRB, American Trucking Association, Alexandria, VA, January 2005.

13. Freight Demand Modeling and Data Improvement Strategic Plan (SHRP2_C20). Transportation Research Board, National Academies of Science, Washington, DC March 20, 2010.

14. "Deepwater Port Plans Put on Ice." Raleigh News and Observer, Thursday, July 22, 2010. Raleigh, NC.