Principles and Guidance for Presenting Active Traffic Management Information to Drivers (2021)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

36 Chapter 3: Literature Review The main objective of this activity was to search through ATM literature and to develop a literature synthesis based on those topics. Figure 40 illustrates an overview of our approach to the literature review. The output of this task was used to identify key research gaps between previous approaches and the research questions from this project. Figure 40: Overview of Literature Review Activities. Methods for the Literature Search To aid the development of our strategy and help to scope our proposed level of effort, we conducted a literature search on ATM topics. The literature search process consisted of two main steps: (a) “broad search,” and (b) “focused search.” The purpose of the “broad search” was to understand research trends on ATM topics and to adjust the search strategy for the second step. The purpose of the “focused search” was to find literature relevant to the 6 key research questions motivating this project. For the broad search, we utilized three resources for the literature search to minimize the possibility of missing key articles due to the constraints of using a single search engine. We used TRID (Transport Research International Documentation), Google Scholar, and MIT library (in that order) as search tools and expanded our reference database by adding new documents. First, we used “Active Traffic Management” as a keyword phrase and found a total of 284 papers from TRID. Second, we repeated this process using Google Scholar, and found a total of 72 papers (among the 72 papers, 27 papers were not covered by the first search, and those were added to our reference database). Finally, we repeated the process using the MIT library15 and found a total of 37 papers. Only four papers among the 37 were not covered by our list, and these four papers were added to the reference database. 15 https://libraries.mit.edu/

37 After the broad search process, two researchers assessed the total of 315 documents through a four-step process: (a) general relevance assessment with respect to ATM, (b) relevance assessment with respect to our research questions, (c) research method classifications (e.g., on- road evaluation, simulation, etc.), and (d) ATM strategy classification. Through these four steps, complete data sources were evaluated to determine if they meet minimum requirements for quality and applicability. Data sources meeting the minimum requirements for quality and applicability were saved for a full review. The results of the general relevance assessment showed that around 23% of the articles that we reviewed were directly related to ATM strategies (44% were not related and 33% were possibly related). We found 36 documents (among the 315) that might be relevant to our research questions (e.g., drivers’ information needs, drivers’ information processing capacity, potential media to deliver information, and effective ways to achieve safe response from drivers), and around 50% of the documents were related to the questions about potential media. However, only a few of the experiments tested alternative/innovative media (e.g., Sykes, 2016). Among the procedures described in the reference database, around 30% of the authors conducted studies related to “VSL,” 17% addressed “shoulder lanes,” and 13% addressed “lane control.” During the broad search process, we found only a few documents that might be relevant to drivers’ information needs and the information processing implications of ATM. Based on the results, we modified our search strategy for the “focused search” process by adding “driver information” as an additional search phrase. For the focused search process, we picked three ATM strategies (“dynamic speed limit,” “shoulder lanes,” and “lane control”), and used these terms as keyword phrases. As our previous results showed a low ratio between the number of documents relevant to our research focus and total number of documents that we found, we combined two keywords—one for ATM strategy and the other for our research focus, “driver information,” and refined the literature search. Another technique that we applied in the focused search was to review Battelle’s previous work, which were generally relevant to the broad topic of “presenting information to drivers, but less relevant to presenting ATM information.” However, some of the questions (e.g., driver information process, information formatting, etc.) can be answered without high relevance with ATM strategies. Especially for question 2 and 4, we have utilized Battelle’s previous work (Campbell et al., 2004; Campbell et al., 2012; Campbell et al., 2016; Lichty et al., 2012; Richard et al., 2015) to find relevant documents to increase search efficiency. From this focused search process, we identified a total of 89 relevant documents. A literature assessment was conducted again for the new additions. After the broad search, focused search, and two rounds of the literature assessment, we had a total of 26 documents representing the final set of relevant data sources. These 26 documents are listed in Appendix B. For the final reference set, the full documents were requested and obtained for further evaluation through the Internet and the Battelle library services. Based on Battelle’s previous experience conducting similar literature reviews during guideline development on a variety of topics, we found that a structured approach to conducting the literature review is essential for developing accurate and unbiased guideline information. A central problem in literature reviews is that the quality and quantity of applicable research findings vary greatly from topic to topic. A

38 systematic approach is therefore needed for the literature review so that only the highest quality and most applicable data available will be used to inform content development of design principles and guidelines. For this reason, we created a review form and reviewer’s guide, which provide reviewers with clear instruction on how to complete the form (see Figure 41 for an example of a completed review). A total of 26 documents were fully reviewed in accordance with the reviewer guide, with relevant information entered into the review form. All the review forms were organized according to research areas or potential guideline topics, and cataloged electronically, allowing for both a project record as well as for future use. This catalog will track the status of individual data sources through the review process and tie the document to the review. Figure 41. Sample document summary template (adapted from McCallum et al., 2006). Literature Synthesis The final step in the literature review process was to develop topic-specific summaries of key findings from the literature, as well as research gaps identified in the reviewed data sources. Table 9 below summarizes key findings related to our research questions (note that Task 3.b focused on research questions 1-4, whereas Task 3.a focused on research questions 3, 5, & 6). Question 1. What information related to ATM strategies does a driver want and need? What characteristics are associated with this information (reliability, timeliness)? From the literature, we have found general characteristics associated with traffic information. Previous studies addressed specific information characteristics and their impact on driver

39 performance and subjective ratings (e.g., Kantowitz et al., 1997), types of general information which drivers need (e.g., Mensonides, 2004), or features of traffic information, such as “accuracy,” “timeliness,” “reliability,” etc. (e.g., Lappin, 2000). However, previous literature did not provide detailed driver needs related to different ATM strategies. As mentioned in the work plan, we aim to understand drivers’ travel/driving-related decisions for specific scenarios related to the key ATM strategies, with details (such as how they make the decisions, what information they needed to make the decisions, where/how they want to obtain that information, etc.). Also, this research question needs to be considered with different ATM mediums (medium type x information type x where/how to obtain). We expect that conducting focus groups will provide specific driver information needs associated with the key ATM strategies. Question 2. How much information can a driver process via the complementary and contrasting modalities (e.g., visual, auditory), given the context and distraction? We have found previous approaches for examining drivers’ information process, range from general legibility and comprehension of the ATM signs (e.g., Jeffers et al., 2015) to effects of in- vehicle alert type/modality on driver behaviors (e.g., Sykes, 2016). In general, “modality comparison” (e.g., visual messages vs. auditory messages) has been studied in the transportation domain, especially for driver-vehicle interface designs. For example, the “Human Factors Design Guidance for Driver-Vehicle Interfaces” (Campbell et al., 2016) describes how to select sensory modalities based on message complexity, receiving locations, information priority, etc., and we expect to utilize these previous guidelines to some extent. What we found as a research gap from the previous literature was that there was little to no systematic comparison among various modalities for ATM strategies. For example, Sykes (2016) compared visual, auditory, and verbal alerts for in-vehicle displays, which disseminated ATM-related information. However, the alert modality was confounded with alert types, and made it unclear whether differences in the outcome measures (e.g., glance behavior) were caused by the modality or alert type (e.g., HOV alert was featured with visual alert, whereas speed limit alert consisted with visual, auditory, and verbal alerts). Based on drivers’ information needs from the research question 1, factorial designs (which systematically manipulate information types/modality) will be applied to evaluate modality effects on driver performances for the specific ATM strategies varying test scenarios. Question 3. What existing and potential media could be used to deliver this information? Media that are under the control of transportation agencies (e.g., electronic signs) are of primary interest but alternative and innovative media (e.g., in-vehicle displays, cell phone applications, geographic information system) and their evolving capabilities and roles must be examined. Potential ATM mediums have been studied by testing alternative options (e.g., Saha et al., 2013; Scarinci et al., 2014; Ishak et al., 2015) or by comparisons with traditional ATM mediums (e.g., Hogema et al., 2000; Craig et al., 2017). In particular, in-vehicle displays have been actively tested as potential mediums in ATM applications (e.g., Sykes, 2016; Hogema et al., 2000). Although we have found more literature related to this topic compared to what we found for the other topic areas, there were only a few studies that directly compared several ATM mediums. In this regard, the Craig et al. (2017) study is noteworthy. Craig et al. (2017) conducted a driving simulator study to evaluate the efficacy of three approaches to presenting a ‘work zone ahead,

40 reduce speed message’ to drivers. Messaging conditions included: (1) an audio message, (2) an audio message + visual message, and (3) a roadside display. Interestingly, the in-vehicle visual message was an image of an icon with text messages, presented within an image of a smart phone. The entire message was presented on an LCD; i.e., the LCD was used to mimic a smart phone. The roadside display was a temporary changeable message sign programmed into the driving route. Overall, the research indicated that the in-vehicle message conditions were associated with better driving performance on key measures, less mental workload, improved perceived usability, and better event recall relative to the roadside display condition. This was perhaps the most directly-relevant study that we reviewed; it demonstrates the potential offered by a shift from infrastructure based ATM messaging to an in-vehicle approach. We had expected that studies employing multiple medium types to disseminate ATM information (e.g., electronic signs and in-vehicle displays) would have been seen in the literature. However, we have not found any literature which considers the harmonization of multiple types of ATM, such as (a) how to coordinate the alternative media (e.g., in-vehicle displays with traditional electronic signs); (b) whether they need to disseminate the same information or need to compensate for each other by minimizing redundancy; (c) how to match ATM information types and delivery medium; or (d) how to maximize the efficiency of information dissemination when deploying multiple ATM mediums. Question 4. Given a particular message and medium, what are effective ways to prioritize, format, and present the information to achieve a desired and safe response by drivers? As mentioned in the work plan, previously Battelle has reviewed original data sources, conducted analytical and empirical studies, and generated entire handbooks to provide effective ways to prioritize, format, and present information to drivers (e.g., Campbell et al., 2012). Conclusions As seen in the State of Practice and Literature Review, relatively little research data is available to support the development of rigorous design guidelines for the six (6) key research questions that animate this project. The state of practice review identified a wide range of deployment for virtually all types of ATM systems. However, these deployments do not reflect findings or conclusions from specific research studies. In general, State DOT staff identified technology and messaging solutions that worked for them, implemented them, and then made changes over time as needed. They generated solutions that worked for their state; often, other states adopted all or part of these solutions. This accounts for the wide variety of practices across the states, yet it provides little in the way of rigorous guidance or even best practices. The literature review identified data sources that provide some very general design guidance for ATM messages. We had hoped to identify data sources/studies more focused on this study’s objectives that could be used to provide answers to key research questions 1-4. However, rigorous answers to key research questions 1-4 were not provided by the existing literature.

41 Table 9: Research synthesis matrix. Kantowitz (1997) Long (2014) Mensonides (2004) Lappin (2000) 1. What information related to ATM strategies does a driver want and need? What characteristics are associated with this information? Information "accuracy" influenced on driver performance and subjective rating for advanced information system. "Dissatisfaction" with VSL signs could be reduced by improving public understanding/operations by modifying current settings. Road users preferred general information about "accidents," "road maintenance," "daily congestion," "events," "slipperiness," "fog." Information features: "accuracy," "timeliness," "reliability," "cost," "degree of decision guidance and personalization," "convenience," "safety.” Janssen (1999) Sykes (2016) Luoma (1999) Jefferes (2015) 2. How much information can a driver process via the complementary and contrasting modalities given the context and distraction? Comparison across six in- vehicle systems (e.g., visual vs. auditory) under car- following and braking tasks. Examination of driver behavior associated with in- vehicle systems by varying alert type and modality (visual, auditory, and verbal). Test whether variable message signs divert driver attention from adjacent fixed signing. Test legibility distance for CMS signs and ATM sign comprehension with static images. Balke (1992) Ishak (2015) Craig (2017) Hogema (2000) 3. What existing and potential media could be used to deliver this information? Changeable message signs, Lane control signals, Highway advisory radio, Commercial radio traffic report. Highway Advisory Radio, DMS, Telephone Information Services. Traditional CMS sign vs. ‘cell phone’ work zone information. Traditional VMS signs vs. in-car speed information. Hourdos (2016) Luoma (1999) Jeffers (2015) 4. Given a particular message and medium, what are effective ways to prioritize, format, and present the information to achieve a desired and safe response by drivers? Specific sign configuration induces more lane changes than others. Drivers pay more attention to highly effective signs using fiber-optic technology than to fixed signs. Comparison of ATM message formats: Washington gantry vs. Minnesota gantry for comprehension and legibility.