Driver Expectations When Navigating Complex Interchanges

Chapter 5. Conclusions and Recommendations

Conclusions

This project yielded several overall conclusions related to driver expectations at complex interchanges. These conclusions are organized around the following two primary objectives of this project:

• How do expectations affect driver behavior?
• What are the related recommendations for sign design?

The findings related to each objective are discussed in the following sections.

How Do Expectations Affect Driver Behavior?

The activities conducted in this project provided information about several aspects of this question. These aspects are discussed in the following sections.

Definition of Driver Expectations

Based on the literature review, it was clear that the concept of driver expectations has not been thoroughly defined with specific regard to interchanges and common driver maneuvers at interchanges. However, this concept has received consideration with regard to broader driving tasks. There are several existing definitions of driver expectations.^(12,3,13) The majority of definitions that were found typically included some variation of the comprehensive definition provided by Lunenfeld and Alexander , which states that expectancy is "a driver's readiness to respond to situations, events, and information in predictable and successful ways."(pg. 153)⁽³⁾ Some of the attributes of driver expectancy include the following:

• Drivers tend to anticipate upcoming situations and events that are common to the road they are traveling.
• Driver expectancy influences response speed and accuracy.
• The more predictable the roadway feature, the less likely will be the chance for errors.
• Drivers experience problems when they are surprised.
• Drivers, in the absence of counter evidence, assume that they will only have to react to standard situations.
• The roadway and its environment upstream of a site create an expectancy of downstream conditions. Drivers experience problems in transition areas and locations with inconsistent design or operation.
• Expectancies are associated with all levels of driving performance and all aspects of the driving situation. This includes expectancies relative to speed, path, direction, the roadway, the environment, geometric design, traffic operations, and traffic control devices.

The empirical findings regarding driver expectations are as follows:

• Drivers expect to need to be in the lane closest to the exiting direction to be able to exit.⁽¹⁷⁾
• Drivers expect lane drops at exits.⁽¹⁸⁾
• Signs and lane markings appear to be effective ways to set driver expectations at interchanges.⁽²⁰⁾
• Driver expectations impact their information acquisition.⁽²¹⁾

Key Driver Expectations Regarding the Navigation of Complex Interchanges

The primary conclusions related to driver expectations come from the focus group discussions described in chapter 3. Some of the conclusions are specific to individual scenarios covered in the focus groups, while others are relevant to more than one scenario. The conclusions include the following:

• Drivers expect that there will be functional relationships between lanes on the roadway and arrows/text on signs and that the signs will make these relationships clear.
• Drivers expect that the distance between a guide sign and a final (or last chance) decision point will be sufficient to allow for making any necessary lane changes in a safe and timely manner.
• Drivers expect that they will have more than one opportunity to obtain necessary destination and lane information before they need to make a final decision regarding lane choices.
• Drivers expect that the freeway system will provide them with the necessary information to construct a mental model and that it will be sufficient to support timely and accurate decisions about lane choice.
• Drivers expect that the information available to them through the freeway system will be sufficient to support decisions about lane choices; at the least, they will never have to move over more than one lane at the last moment.
• Drivers expect that the freeway system will provide sufficient information to support decisions about all route choices, not just frequent or popular choices.

How Various Interchange Elements Affect Driver Tasks and Performance

This question was primarily addressed in the task analysis described in an earlier project report.⁽⁴⁾ This analysis provided high-level information about how some interchange elements have the potential to affect driving, and the relevant findings are briefly summarized for various interchange elements as follows:

• Curves/ramps: These represent one of the most significant sources of elevated driver workload because these elements inherently involve a greater number of driving tasks, and they contain more time-based restrictions on task performance and decisions.
• Guide signs: These represent a key source of driver visual and cognitive workload, and they can make the cognitive navigation task unnecessarily complicated depending on what information is presented and how it is laid out on the sign. The spacing and placement of signs (particularly with sight-distance restrictions) is also a key determinant of workload.
• Merge/weave sections: These sections impose visual and psychomotor demands on drivers as they manage their interactions with other traffic.
• Option lanes: Signs that indicate option lane information using down arrows may lead to unnecessary effort because they require drivers to integrate disparate information and deduce the intended meaning. Also, diagrammatic signs rely on relatively small sign elements (lane marking symbols) to communicate option lane information, which may be difficult for drivers (especially older drivers) to see at fast speeds. This may cause them to miss this important information.

Key Challenges to Drivers

The task analysis also indicated that drivers face several key challenges in interchange driving. These challenges include the following:⁽⁴⁾

• Multiple concurrent visual activities: Typically, on ramps and curves, drivers face multiple concurrent visual tasks that likely involve switching between targets, and they require that drivers manage where they deploy their glances, which increases workload. Drivers may also be forced to only read part of a sign or make decisions with less thought and deliberation than they may prefer. As the task 4 focus groups indicated, these types of rushed actions make drivers uncomfortable and stressed out about interchange driving.
• Time-limited nature of interchange driving: A key problem for drivers is having sufficient time to make sound decisions in some of the more time-limited situations. In particular, some decisions must be made close in time to other tasks that can interfere with decisionmaking, such as reading signs and managing traffic interactions. Another challenge is the time limitations imposed by interchange geometry, such as sight-distance restrictions and lane drops/gore points. These elements are fixed in time and space for drivers, and they place hard limits on the time available to conduct intervening activities. While drivers always have the option of slowing down to give themselves more time, this is yet another decision. Also, as the task 4 focus groups indicate, many drivers feel social pressure to keep with the flow of traffic. Several drivers reported that they would rather miss their exit and double-back when faced with time-pressure conditions.
• Managing and interpreting navigation information: Driver navigation expectations for less common destinations may be significantly more complicated to understand. More specifically, drivers have to keep track of more information to figure out where to go in absence of direct navigation information. This not only makes drivers work harder but makes it more likely that drivers will make incorrect decisions. The task 4 focus groups indicated that drivers expect signs to provide sufficient information to support decisions about all route choices, not just frequent or popular destinations.

Safety and Capacity Implications of Driver Expectations Findings

The findings from this project illustrate some of the implications of driver expectations on safety and capacity at complex interchanges. Effects related to safety and capacity are discussed in the following subsections. It should be noted that this information is primarily based on qualitative focus group discussions or empirical data from a limited set of drivers. Therefore, these findings should be viewed as suggestive of the driver expectations that could impact safety or capacity, but their link to actual traffic data is not confirmed in any way in this project.

Safety:

The concerns related to safety generally involve drivers at complex interchanges having to execute multiple actions with limited time, which can lead to drivers making more mistakes in situations that generally have lower margins for error. Some of the specific safety-related implications are as follows:

• Multiple lane changes in a short distance: Drivers reported their previous experiences with other drivers making multiple last-second lane changes to reach exits, and some reported that they would do so themselves. This could lead to drivers taking excessive risks while changing lanes or executing them before fully preparing (e.g., foregoing shoulder checks).
• Time-limited actions: According to the task analysis results and driver comments about the focus group scenarios, the time available to read certain guide signs can be limited by sight or reading distance. This limits the subsequent time available to make navigation decisions or execute maneuvers such as lane changes. Rushing these activities could promote driver errors that lead to unsafe conditions (e.g., erratic driving to complete a maneuver in time).
• Concurrent driver tasks: The task analysis indicated that drivers can sometimes encounter increased activity levels on ramps such as changing speed, steering, reading signs, and making lane changes. There is potential for interference across these various activities (e.g., drivers reading a sign could lose vehicle control on curved exit ramps).

Capacity:

Most of the implications regarding interchange capacity involve uneven lane usage; however, effects related to drivers slowing or missing exits are as follows:

• Lane usage: Several of the results from the empirical data collection suggested that if drivers have some uncertainty about whether the option lane serves their destination (e.g., due to ambiguous destination information on guide signs) or if they are unsure that it is an option lane (e.g., poor alignment of guide sign arrows with lanes on a curved segment), they will strongly prefer getting into the exit only lane over using the option lane. The converse is also true. Specifically, if drivers have any doubts that an exit only lane will lead to their destination, they will try to be in the option lane to avoid mistakenly exiting the freeway. This is consistent with a general trend observed in the focus group comments in which drivers reported being conservative with their lane choices when faced with uncertainty about which lane serves their movement. In these cases, drivers prefer to avoid outside lanes so that they have more options and can avoid making multiple lane changes in a short distance.
  

  It is worth noting that the described lane usage patterns pertain mostly to drivers who are unfamiliar with the interchange. There were focus group comments suggesting that familiar drivers actually improve capacity by avoiding congested lanes when they know that other lanes will also take them to their destination. An extension of this notion is that if drivers clearly understand which lane goes where, they may make better use of less crowded lane alternatives. Therefore, if it is possible to communicate lane information sufficiently using guide signs or other approaches, even unfamiliar drivers could act as familiar drivers and optimize capacity.

• Other impacts on capacity: The project identified other ways in which capacity can be negatively affected by driver actions in response to complex guide sign information. As alluded to in the previous paragraph, the strategies that drivers adopt for dealing with uncertainty involves making additional lane changes, many of which may be unnecessary. Another common driver action that can reduce capacity is that drivers in the focus groups reported slowing down to read signs or to give themselves more time to make decisions or change lanes. Finally, a less direct impact on capacity is related to drivers missing exits because they could not complete the movement in time or because they felt that it was unsafe to execute a movement. Drivers commonly reported that they would exit and re-enter the freeway to try to make their intended exit, which unnecessarily increases traffic flow, although by a small amount.

Recommendations for Sign Design

The following sections summarize the key conclusions and findings regarding recommendations for sign design that were identified in this project.

Key Design Principles and Guidance from Existing Research

One outcome of the task 2 literature review was an initial set of key design principles, which include the following:

1. Provide adequate forward sight distance.
2. Provide transition cues.
3. Minimize attention-dividing conditions.
4. Provide navigation information to address all of the driver information needs.
5. Maintain compatibility between the interchange and the visual cues.
6. Design to accommodate the drivers' expectations and abilities.
7. Warn drivers of situations which may violate their expectations.
8. Allow drivers to recover after making an error.
9. Design for simplicity.
10. Design for consistency and predictability.

The "Results" section in chapter 2 provides additional details about specific steps that can be taken to address each design principle.

Findings Related to Sign Design from the Task 7 Data Collection Activities

The task 7 empirical activities yielded several findings that support the development of specific recommendations for guide sign design. The spatial organization and layout of guide sign information have an important influence on driver interpretation of signs and their expectations of upcoming interchange geometry. The clearest and most consistent finding from task 7 is that perceptual factors related to the organization of information on a sign influence how drivers interpret the sign. The focus group discussions also provided similar findings. These perceptual factors can involve the layout of informational elements, how sign elements are grouped, and/or the position of the sign on the sign bridge. The data suggest that some drivers formed strong expectations based on the arrangement of sign information in certain situations. This is especially apparent in topic 3 of task 7 in which sign placement on the bridge was a stronger cue than the specific sign notation for communicating a left exit. Also, there is evidence from this same topic that drivers confused left exit panels with exit only panels when those panels were located on the right side of the sign bridge. This suggests that the expectations that drivers had related to sign position biased their understanding of the sign. Under time-constrained driving conditions, these expectations can influence their reading of signs and make the difference between correctly or incorrectly interpreting a guide sign. Other specific findings related to the influence of perceptual factors are discussed in the "Conclusions" section of chapter 4.

In the focus groups, most drivers appeared to interpret down lane arrows rather narrowly to correspond to the destination that is directly above or grouped with the arrow. For example, in scenario 1, while both NW Ind. Area and St. Helens destinations could be reached by either of the two right lanes, most drivers paired each destination with only the lane directly below the corresponding lane arrow. This tendency led to problems later in the scenario because drivers associated the rightmost lane, which was an exit only lane, with St. Helens when the exit was for a different destination (Vaughn St.).

Understanding the perceptual factors that influence guide sign interpretation is important because they represent attributes that can be exploited to make signs more useful to drivers. However, if signs are designed without proper consideration of these factors, it could also lead to unnecessarily complicated or confusing guide signs. This is especially important for complex interchanges because signs typically communicate more information, the interchanges are more likely to be unique and unfamiliar to drivers, and the interchanges are uncommon and are more likely to involve atypical geometric elements.

Up arrows yield consistently better results than down arrows in terms of accurate driver understanding of permissible interchange movements and efficient option lane usage. In topics 1, 2, and 4, up arrows led to more efficient option lane usage or better comprehension of permissible movements when compared to down arrows. Up arrows differ from down arrows in multiple ways. They often have a clearer visual alignment with a single roadway lane. The movement information can be inherent in the arrow (e.g., curved to the right for a right exit or straight for the through destination). Additionally, since the curvature of the arrow can be varied by destination direction, up arrows can be grouped with other up arrows pointing the same direction (e.g., to indicate that both an option lane and an exit only lane lead to the same destination). Overall, these findings suggest that drivers require additional or specific information about option lanes and the movements they represent and that up arrows can be used more effectively to provide this option lane information.

An aspect of sign comprehension that was examined in this study was how drivers visually group destination information with other sign elements and particular roadway lanes. A key sign element used for communicating which information goes together is the type of separator used to divide destination labels on the sign. Three separators were investigated in this study: vertical lines, hyphens, and multiline separations. In addition to these, data suggest that the presence of an exit only panel (more specifically, the space it occupied along the bottom of the sign) may also have acted as a visual separator. The results found for the various separators are as follows:

• Vertical lines caused drivers to associate the destination on either side of the line with only the lane below the destination label. This separator appeared to supersede the effects of an exit only panel that spanned the entire sign width.
• Hyphen separators had mixed results. When used with an exit only panel that spanned the entire sign width, hyphen separators were predominantly interpreted as indicating that both lanes go to both destinations. When used with an exit only panel that only spanned the rightmost of two exiting lanes, the hyphen separator caused some drivers to associate each destination with only the lane immediately below it.
• Multiline separators did not function as a separator at all; rather, they led drivers to group both of the destinations with both of the lanes that they were centered above. This type of grouping was best for pairing multiple destinations with the same information. This happened regardless of the style of exit only panel used.

Clearly, separation cues influence how drivers associate destination information with specific lanes. Although these elements were not tested exhaustively or necessarily in isolation, these findings demonstrate the importance of the choice of separator, as each led to different expectations for the upcoming roadway, particularly when used in conjunction with an exit only panel. Additionally, the exit only panel itself deserves consideration as a separator since it appeared to influence how drivers grouped destination information with lanes.

Recommendations

Based on the work conducted in this project, there are several recommendations for follow-up activities that can help researchers understand driver expectations at complex interchanges. Follow-up activities related to the task analysis are provided in the task 5 report.⁽⁴⁾ Follow-up activities related to task 7 data collection are described in this section.

Additional research should be conducted to obtain a more complete understanding of the influence of perceptual factors on guide sign interpretation. The current research clearly demonstrates that perceptual factors exert a strong influence on driver interpretation and responses to guide sign information. The first conclusion in the previous section discusses the significance of this finding in detail. However, while the data collected in task 7 provides an initial appreciation of the importance of perceptual factors on driver expectations and their interpretation of guide signs, the resulting data represent an initial look at some of these factors. A more comprehensive and systematic approach is required to fully understand the role of information layout and organization on driver comprehension and interpretation of guide signs.

As mentioned previously, a key benefit of understanding the perceptual factors that influence guide sign interpretation is that they represent attributes that can be exploited to make signs more useful to drivers. Also, failing to properly consider these factors could lead to unnecessarily complicated or confusing guide signs. There are other useful applications for this type of information that a more detailed investigation of this topic would provide as follows:

• Developing recommendations for sign design and layout: Currently, most of these perceptual elements are not considered in the design process, and they are not typically directly addressed in the MUTCD.⁽¹⁾ New sign designs require engineers to make many decisions regarding placing information elements, grouping elements, arrow types, etc., particularly when creating a novel sign for a new, complex interchange. Specific information about the consequences of element usage can assist informed decisions for these new sign designs, and it can be incorporated into existing sign design references.
• Identifying cost effective sign design trade-offs for new or refurbished guide signs: Similar to the previous point, each sign element has an associated production cost, which may be factored into the design decisions. When considering element-level trade-offs, it is important to know the consequences of selecting a particular element in terms of driver behavior to determine if a low-cost compromise is viable.
• Diagnosing problem interchanges: In situations where traffic engineers are noticing a disparity in the distribution of lane usage, lane changes close to the gore point, or unnecessary slowing, it may be possible to use information about driver comprehension of sign elements as a diagnostic tool to specifically determine which elements or groupings are causing driver confusion.

Overall, the importance of perceptual factors in guide sign design is a relatively unexplored topic, yet the data from the current research indicates that it is potentially a key topic for designing guide signs that can be quickly and accurately understood by drivers. Conducting additional research in this area will support activities that will improve guide sign design and facilitate driver navigation of complex interchanges.

Systematically Apply Methodologies to Additional Interchanges with Existing Navigation Challenges

This project made good progress in developing methods and tools for identifying driver expectations at complex interchanges and applying the findings to generate solutions for problems that drivers encounter. A logical extension of this work is to apply this process to additional complex interchanges that are known to cause problems for drivers. Using the approaches from this project in a systematic way would also be very useful for refining these methods, developing tools that engineers could use to identify specific problems at complex interchanges, and then finding solutions for addressing them. An example of a geometry that could be evaluated is interchanges that include collector-distributor roadways. These roadways would be valuable to investigate since there is little guidance available regarding designing for driver expectations there, and they can be complicated for drivers to navigate (e.g., a single exit may need to include signage for a complex configuration of multiple future exits or interchanges that it serves).

To extend the current methodology to examine other interchanges, the following steps would need to be taken:

1. Identify candidate complex interchanges with potential navigation issues that could be addressed. Once the target interchanges have been identified, it would be necessary to work with State transportation departments and other stakeholders to understand their design challenges/constraints and identify information needs regarding driver navigation.
2. Conduct focus groups, task analyses, and empirical data collection to obtain information about driver expectations, specific stressors or challenges encountered, information needs, and workload/task requirements at the selected interchanges. This information will be used to develop signing strategies and specific guide sign designs that meet driver information needs for the tested interchanges. These strategies and designs could then be evaluated at key points along the roadway and for specific movements with a new driver sample. This would involve measuring driver comprehension, driver assumptions and expectations, and resulting actions (e.g., lane choices) using data collection approaches developed in the current project.
3. Develop specific recommendations for improving signing at the interchanges examined and more general recommendations that can be applied to related systems.

A key outcome from this research would be specific recommendations for signing plans that are validated by driver testing for each interchange investigated in the project. This would include specific guide sign designs that were shown to best support driver navigation and expectations. More general recommendations would be developed for related interchanges, which would be turned into a guidance document to inform new designs or improvements at other related interchanges. Another outcome would be to refine the approaches and tools developed in this current driver expectations project to make them more directly applicable and useful for solving a broader set of signing design challenges.

Conduct Research to Understand Why Up Arrows are More Effective at Communicating Permissible Lane Information

In several topics examined in this study, up arrows outperformed down arrows in aspects of lane usage or communicating lane assignments. This leads to the question: which characteristics of up arrows make them more effective? Although up arrows are conceptually similar to down arrows, it appears that there may be a basic difference between them that causes drivers to perceive them differently. For example, if drivers are reading the arrows from base to point, down arrows appear to be focused on assigning destinations to lanes, while up arrows start at the driver and point to where they are going. Similarly, because up arrows can show multiple curvatures or directional combinations, they can be used to communicate additional information to drivers, often in an implicit manner (e.g., using the same curvature to visually associate movements in separate lanes that reach the same destination).

Conducting research to obtain a clear understanding of how drivers interpret different aspects of up arrows could provide additional tools for communicating interchange movements in a manner that can be immediately understood by drivers. This is especially important at complex interchanges because arrows are often used in unconventional ways to communicate novel, complex information (e.g., sign sets B and B2 in topic 7).

The new information from the current project can be used to update the Human Factors Guidelines for Road Systems.⁽⁵⁾ The publication has an entire chapter of guidelines dedicated to interchanges; however, information about the effects of various sign elements on driver expectations in complex interchanges is clearly lacking.⁽⁵⁾ It would be valuable for roadway designers to have more information about the consequences of using certain sign elements over others in terms of the effects on driver understanding and lane selection. An immediate contribution of the current study could be to fill some of the knowledge gaps about these issues. Some information from the current study that could be applied to the Human Factors Guidelines for Road Systems includes the following:⁽⁵⁾

• There is a limit on the number of lanes that can be depicted and successfully comprehended on a diagrammatic guide sign (see topic 6). This limit is even smaller for signs with an equal number of lanes in each branch of a split than an unequal number of lanes in each branch.
• Complex signs for depicting the upcoming geometry can assist driver understanding (see topic 7). However, each destination is impacted by the individual elements used, and it is important to consider every possible driver movement.
• Overall, individual sign elements impact driver grouping and lane assignments for each destination. Although many aspects of sign design are regulated by the MUTCD , when designing a novel sign, it is important to consider the assigned meaning of each element that is used because drivers assign meaning to each element that they see.⁽¹⁾

Conduct a More Comprehensive Analysis of Individual Response Patterns

This recommendation originates from the discussion on variability in driver interpretation of guide signs. The task 7 response booklet was used to collect demographic information about driving history and experience with interchanges; however, neither this information nor information about driver demographics, such as age, were included as factors in the analyses. A simple follow-up activity for this project would be to take a closer look at the importance of these factors as predictors of driver responses. For example, professional drivers may have more effective strategies for focusing on key elements of signs while ignoring less relevant information. If this pattern was observed in the new analysis, then it could lead to sign designs that better promote the identification of key information over secondary information. There are also important questions that can be investigated, including the following:

• Are there groups of drivers who consistently interpret signs the same way?
• Are these groups defined by demographics, driving experience, or some other factor?
• At the level of the individual, how stable are the views that define the groups (i.e., drivers responding the same way every time they see a particular sign element versus supplying responses that are situation-specific)?

The results from the current investigation show that there is still a substantial degree of unaccounted variance in driver responses. Obtaining a better understanding of this variance can provide additional information that can support the design of effective interchange guide signs.

Another longer-term approach for furthering researcher understanding of driver expectations at complex interchanges is to develop work plans for investigating these questions using SHRP2 naturalistic driving data. The SHRP2 naturalistic driving study is a large, on-road data collection effort that involves instrumenting approximately 3,000 drivers' own vehicles with an advanced data collection system including internal and external viewing cameras, forward radar, Global Positioning Systems (GPSs), and other sensors. Data collection is ongoing until the end of 2013, with data becoming available afterward. An important and unique aspect of SHRP2 data is that it will record a driver's behavior over a potentially large number of repeated traversals of the same interchanges. Using these data, it should be possible to develop baseline interchange navigation profiles for specific drivers using data from repeated movements over frequently traversed simple and complex interchanges. For example, one aspect of a driver's profile could reflect the degree to which that individual prefers the option lane versus an exit only lane. It might also be possible to identify factors that lead to changes in this behavior (e.g., a driver who typically prefers an exit only lane may choose the option if traffic is heavy). It should also be possible to examine the same driver at complex interchanges that they rarely traverse to determine how behavior in a comparable, unfamiliar complex interchange differs from the baseline profile. Using this approach, it may be possible to correlate features of the complex interchanges (e.g., geometry, sign information, etc.) with changes in behavior relative to baseline performance. This activity could provide information about how features of complex interchanges impacted driving behavior, and these findings could be used to provide behavioral validation of the types of findings observed using static and driving simulator data collection approaches.

It might also be possible to examine driver behavior at unfamiliar complex interchanges at a finer level of detail. For example, in-vehicle video data could provide information about when drivers look at signs and how much time they spend reading them. Other vehicle data can provide information about changes in vehicle speed as drivers prepare for movements as well as measurements of the distances at which they try to initiate lane changes. The data can also provide indications of navigation errors in the form of GPS traces that show interchange reentry and exit at different locations. Overall, the SHRP2 naturalistic driving data provide a rich and comprehensive dataset for expanding this initial work to obtain a better understanding of driver expectations at complex interchanges.