7 Fast Externally-paced Navigation – Memory and Action Selection in Human-Machine Interaction

7
Fast Externally-paced Navigation

With the focus of action selection processes involved in fast externally-paced navigation, this chapter illustrates a case study that adopted the CCE methodology to derive information necessary for safe and enjoyable driving for drivers while driving. Note that this is a time-critical situation where System 2 before mode and System 1 before mode should be appropriately coordinated along the information provided externally, e.g. through a car navigation system of a human navigator. The study was designed in such a way that participants as drivers should perform event-based navigation, and have aided to conduct otherwise impossible driving in System 2 before mode, i.e. anticipation-based navigation, by “timely” provided “appropriate” information.

7.1. Introduction1

The primary purpose of driving is to transfer persons or things from one place to another. The primary mission of the driver is to control the vehicle safely. This may include the achievement of objectives such as energy efficiency, punctuality and so forth, as optional missions. Traditional car navigation systems are expected to provide information that helps drivers to accomplish their mission of maneuvering satisfactorily. However, with the advent of telematics, modern car navigation systems have begun providing location-based information, such as recommendations of local restaurants along the route. The infrastructure enabling the provision of location-based and context-sensitive information is being developed, making it possible to provide information to drivers that would allow them to experience more safe and enjoyable driving in terms of not only the maneuvering mission but also a variety of concerns encountered en route.

Technological development is very rapid, and it will be possible in the near future to provide information that is necessary for drivers in order to make driving safe and enjoyable. However, what is lacking at the moment is knowledge concerning what information is actually needed by drivers. Due to this problem, the provision of information by the modern car navigation systems is not completely appropriate. In other words, “know the drivers” is the key issue. “Drivers” are considered as “users” to be understood in this chapter. In some cases, the content of the provided information is not what the drivers actually need, and therefore it may be evaluated as annoying. In other cases, the information is not provided in a form that the driver can easily understand, e.g. the driver may prefer “distance to the intersection to turn” over “the number of blocks to the intersection to turn”.

This chapter describes a series of on-road observations conducted as a CCE study for elucidating the nature of information that drivers consider necessary to make driving safe and enjoyable. This knowledge should provide a basis for designing an information-provision service for safe and enjoyable driving. A list of such information based on the results of the on-road observations are identified. Some kinds of information can be easily provided by applying currently available or developing technology, whereas other kinds will require extensive technological development.

Similar to the procedure discussed in Chapter 6 and according to the CCE steps as shown in Figure 5.1 (Chapter 5), the study was conducted as follows:

  1. Step 1: navigational characteristics similar to the characteristics discussed in Chapter 6 were obtained from the field observations, i.e. such cognitive functions as planning for foreseeing future driving routes and road conditions, attention for selectively focusing on objects necessary for safety driving in the environment, and working memory for keeping the task relevant information active for performing safe and enjoyable driving.
  2. Steps 2 and 3: we conducted simulations by mapping these cognitive functions on MHP/RT as shown in Figure 2.10, and derived ideas for a field study; three kinds of information for safe and enjoyable driving were identified:
    1. 1) guidance for routing;
    2. 2) support for safe driving;
    3. 3) provision of miscellaneous information, such as information about daily topics of interest to the driver and information about interesting things to see along the route.
  3. Step 4: then, we designed a CCE study in which four pairs of participants were chosen from among those responding to a Web survey and attending a follow-up interview.
  4. Step 5: each pair was asked to drive along six routes. Three of the routes were familiar to one of the pair and new to the other, with the former serving as navigator and the latter serving as driver. For the other three routes, the roles were reversed. Three interviews were conducted, one coming after two drives in which the pair played both roles in order to derive information considered necessary for safe and enjoyable driving by the participants who served as drivers on the routes unknown to them.

7.2. Steps 1 and 2 of CCE

Navigation means “going from one place to another”, where “place” could be either a physical place in the four-dimensional universe or a virtual place in informational space. Navigation could be either fast or slow in comparison with the characteristic times of human action as specified by Newell’s time scale of action (Figure 2.11). As mentioned in Chapter 6, navigation speed is at the order of walking speed, i.e. 4–5 km/hr, whereas in this chapter, it is about at the order of driving speed, i.e. 30–100 km/hr. However, the analysis described in section 6.2 is still valid even if the characteristic speed is faster for driving. Therefore, even in the case of driving, the following cognitive functions described in section 6.2.1 are considered as important while driving:

  1. – attention;
  2. – planning;
  3. – working memory.

An MHP/RT simulation of driver behavior is similar to the one mentioned in section 6.2.2 because the same cognitive functions are assumed to affect behavior of drivers. Therefore, the following four components of MHP/RT should have effect on the detailed performance of drivers:

  1. – sensory information filter;
  2. – resonance;
  3. – feedback from consciousness;
  4. – frame rate.

7.2.1. Designing a CCE study

The focus of this study is how the processes described in Figure 2.10 should become smoother when some information is provided to the driver who cannot otherwise deal with the driving situation smoothly. In other words, identification of conditions of externally provided information that goes through sensory information filter and resonate with the driver’s memory to help create appropriate feedback from consciousness within the range of frame rate of update of working memory, which is used for other on-going processes for driving. The three cognitive functions need to be coordinated appropriately in the ever-changing driving environment when such information is provided in context:

  1. 1) guidance for routing;
  2. 2) support for safe driving;
  3. 3) provision of miscellaneous information, such as information about daily topics of interest to the driver and information about interesting things to see along the route.

This requires special considerations when designing a CCE study, which is very different from the one described in Chapter 6 for slow self-paced navigation, where little coordination among cognitive functions was necessary, or timeliness and appropriateness of external information should not have had large effects on the passengers’ behavior.

7.2.1.1. Outline of the study

We need to conduct a series of on-road observations to extract information necessary for safe and enjoyable driving. A number of pairs of participants, say four pairs, join the study. The two persons in each pair should be acquainted with each other very well and know each other’s driving attitude and knowledge about driving, so they are expected to provide information necessary for safe and enjoyable driving to the driver when their partner is driving along an unfamiliar route.

Each pair will participate in three sessions with different purposes. Each session consists of a set of two on-road drives with driving times ranging from 30 to 90 min, followed by an interview. The purpose of the interview is to determine the information needed for safe and enjoyable driving. Each drive is videotaped using two digital video cameras as shown in Figure 7.1: one for the outside view (camera B), and the other for the driver and navigator view (camera A).

Figure 7.1. Experimental setup (adapted from [KIT 09])

7.2.1.1.1. First session: know each other

In the first drive of the first session, one member of the pair drives a route that his/her partner is completely familiar with. The partner thus serves as a human navigator who is expected to provide information that will help the driver to make the journey safe and enjoyable. In the second drive in the first session, the roles are reversed, that is the person who drives in the first drive serves as the navigator, and vice versa. The navigator is expected to provide information necessary for safe and enjoyable driving to the partner, i.e. the driver, by imagining that he or she, the navigator, is driving his/her familiar route and estimating the degree of necessity of provision of information to the driver who is not familiar with the route using his or her own criteria. However, since the navigator does not know exactly what information is actually needed by the driver, some information would not be perceived as necessary by the driver. Therefore, the purpose of the first session is to “know each other better in order to provide information that is actually necessary for safe and enjoyable driving”. An interview session conducted immediately after the two drives is designed to facilitate this “know each other” learning. However, the learning is case-based. Thus, the acquired knowledge should not be general enough to be applied to similar situations that do not exactly match the learned knowledge.

7.2.1.1.2. Second session: use case-based knowledge for providing useful information

As the result of the first session, the navigators are expected to have acquired case-based knowledge as to which information should have been perceived by the drivers as necessary for safe and enjoyable driving. The second session asks the navigators to utilize this set of knowledge. As per the first session, one participant serves as a driver in the first drive in the second session and as a navigator in the second drive in the second session, another vice versa. In the interview session, after two drives, their driving behavior is reviewed. This experience should help the navigators to strengthen their case-based knowledge that has been acquired in the first session by using it in the second session. In addition, during the interview session, the navigators are provided with a list of generalized information that are derived by analyzing the instances in the drives in the first and second sessions when information is needed for safe and enjoyable driving. For example, a concrete instruction “turn right at the second traffic light” could be generalized to an expression such as “use traffic lights or landmarks when instructing”.

7.2.1.1.3. Third session: use general knowledge for providing useful information

As the result of the second session, the lists of generalized information are handed to the navigators in order to make sure that the listed information would be provided to the drivers when necessary. Two drives follow for each pair of participants exactly same as the first and second sessions.

7.3. Step 3 of CCE: monitor recruiting

The study was conducted from October 2007 to February 2008. It was crucial for this study to recruit participants who were good at providing information for safe and enjoyable driving to the driver. In other words, a person, serving as a navigator for his/her familiar route, has to be able to:

  1. 1) simulate a driver’s mental processes while he/she is driving an unfamiliar route;
  2. 2) find situations where the driver may have some difficulty in deciding what to do while driving;
  3. 3) compose appropriate guidance for him/her considering his/her preference to the way in which it is provided;
  4. 4) provide it in a timely manner considering the driving and traffic situations.

Therefore, we adopted a two-stage recruiting process consisting of a Web questionnaire for screening and an interview for the final selection.

7.3.1. Screening

Questionnaires were distributed to 1,655 individuals who met the following conditions:

  1. 1) He/she lives in the metropolitan area of Tokyo;
  2. 2) His/her age is between 20 and 50 years;
  3. 3) He/she drives his/her car more than twice a month.

Some questions requested them to self-estimate their skill in providing the information necessary for safe and enjoyable driving, such as:

  1. – are you an officious person?
  2. – are you an industrious person?
  3. – are you a considerate person?

A respondent was asked to accompany a person who would participate in the driving sessions. He/she answered these questions on behalf of his/her partner.

The responses were analyzed to select candidate pairs for the on-road observational study. The judgment criterion was whether both the respondent and his/her partner would be able to provide information that was necessary for safe and enjoyable driving while one of them was navigating and the other was driving. Note that both members of the pair would play the roles of navigator and driver. Thus, each was expected to be a good information provider for the other. Ten pairs were selected as candidates for the on-road observations, then further examined in an interview, described below, for the final selection.

7.3.2. Interview for the final selection

There were several stages in the interview. First, the answers provided by the respondents were reviewed for their appropriateness. Each candidate pair was then asked to plan the routes for the three driving sessions. A total of six routes were provided. A requirement for each route was that the route was completely familiar to the navigator and yet was new to the driver. In addition, it was anticipated that there would be a number of opportunities for the navigator to provide information necessary for safe and enjoyable driving. During the interview, the candidate pairs were asked to draw the routes on a board. The drawings were used to examine the appropriateness of the routes for this study with a face-to-face interview. An example drawing is shown in Figure 7.2.

Figure 7.2. An example of route sketched by a participant

Following the examination of the responses of the candidate pairs, four pairs were selected for the on-road observational study. Three of the four pairs were married couples, and one was a pair of female colleagues who worked at the same company.

7.4. Steps 4 and 5 of CCE: monitor behavior observation and individual model construction

7.4.1. Data analysis

A total of 24 maneuvers were videotaped and analyzed in order to derive the information provided by the navigators to the drivers. As a result, a total of 1,859 information sets were extracted.

7.4.1.1. Data coding

Each information set was represented using the following four descriptors:

  1. 1) The utterance;
  2. 2) The way the information was provided including the timing, and the way the direction was described;
  3. 3) The traffic conditions, such as the width of road, speed of the vehicle, etc.;
  4. 4) The driver’s condition, such as the workload of driving or any preexisting knowledge of the route. The former was estimated by viewing the video and the latter was inferred during the interviews.

7.4.1.2. Evaluation of the provided information

In the interviews, each information set was evaluated by the driver as to whether it was effective in helping him/her to conduct safe and enjoyable driving. In addition, likely rationales for the evaluations were attached when possible. In other words, the information provided by the navigator, who judged it should be helpful for the driver based on the result of his/her simulation of driver’s mental processes at the time of provision of the information, was examined whether it had been as such by the driver. The accuracy of the simulation was examined and revised if necessary.

However, in reality, it was not possible to allocate enough time to review all of the information, so parts of the provided information were not reviewed thoroughly. We evaluated such information by ranking them at one of the following levels with the help of “simulation model of the driver” as we derived from the interview session:

  1. 1) The information was judged as good because it seemed necessary for the driver to conduct safe and enjoyable driving.
  2. 2) The information was judged as average because it was not likely to have affected the driver’s behavior.
  3. 3) The information was judged as poor because it seemed that the driver responded negatively.

7.4.2. Information necessary for safe and enjoyable driving

Items evaluated as good were extracted from the whole set of provided information and further classified into three categories:

  1. 1) Guidance for routing.
  2. 2) Support for safe driving.
  3. 3) Provision of miscellaneous information.

In the following, the subcategories of these categories will be described in detail.

7.4.2.1. Category 1: guidance for routing

There are nine subcategories under this category as follows:

  1. 1) Information that specified the point of action by using an easy-to-understand references for the driver (see section 7.4.3.1 for a detailed example).
  2. 2) Information that was useful in planning future actions.
  3. 3) Information whose expression was crafted by selecting words that were comprehensible to the driver.
  4. 4) Information that augmented the driver’s knowledge about the route.
  5. 5) Information that specified a route that the driver preferred.
  6. 6) Suggestions of drop-by places made by considering the driver’s condition.
  7. 7) Suggestions of parking places that made the activities undertaken after getting out of the car easier.
  8. 8) Timely provision of information that made the driver feel comfortable and relieved.
  9. 9) Information that confirmed an action that the driver had carried out according to the navigator’s directions.

7.4.2.2. Category 2: support for safe driving

There are seven subcategories under this category as follows:

  1. 1) Information that helped the driver to resolve undecided situations, leading to safe driving.
  2. 2) Information that caused the driver to pay attention to events for safe driving (see section 7.4.3.2 for a detailed example).
  3. 3) Information that helped the driver prepare for future actions, ensuring safe and smooth driving (see section 7.4.3.3 for a detailed example).
  4. 4) Actions carried out by the navigator that otherwise would have had to have been carried out by the driver.
  5. 5) Checks for traffic conditions carried out by the navigator that otherwise would have had to have been carried out by the driver.
  6. 6) Pointing out actions missed by the driver.
  7. 7) Reflections on improvements in driving behavior.

7.4.2.3. Category 3: provision of miscellaneous information

There are two subcategories under this category as follows:

  1. 1) Information about daily topics of interest to the driver.
  2. 2) Information about interesting things seen along the route (see section 7.4.3.4 for a detailed example).

7.4.3. Examples

In this section, four episodes involving the provision of information that was perceived by the driver as necessary information for safe and enjoyable driving will be demonstrated.

7.4.3.1. Episode 1

Information which specified the point of action by using an easy-to-understand reference for the driver (see Figure 7.3) was estimated as helpful information. When there was a leading vehicle, the navigator provided routing information by using the leading vehicle as the reference. The area where this guidance was provided had a number of traffic lights and intersections. Therefore, it was not easy for the first-time driver to identify the exact location to make a turn. Thus, this information was evaluated as highly valuable by the driver.

Figure 7.3. Episode 1: information that specified the point of action by using an easy-to-understand reference for the driver (adapted from [KIT 09])

7.4.3.2. Episode 2

Information which caused the driver to pay attention to events for safe driving (Figure 7.4) was considered helpful. The navigator told the driver that he/she should pay attention to motorcycles that would likely pass on the driver’s side of the car. This area was near a university campus and many students had motorcycles.

7.4.3.3. Episode 3

Information helped the driver prepare for future actions, ensuring safe and smooth driving (Figure 7.5). When the driver was about to make a lane change from the current driver-side lane to the passenger-side lane in order to prepare for the next left turn, the navigator suggested remaining in the driver-side lane until the car had passed a mass home electronics retailer because it was likely that the passenger-side lane would be jammed with cars going to the retailer.

7.4.3.4. Episode 4

Information about interesting things seen along the route was helpful for driver. The navigator introduced the driver to a bakery where a variety of excellent breads were sold at reasonable prices. The navigator knew that the driver was interested in bread and that this information would be highly valuable for her/him.

Figure 7.4. Episode 2: information that caused the driver to pay attention to events for safe driving (adapted from [KIT 09])

Figure 7.5. Episode 3: information for safe and smooth driving. Note that cars drive on the left lanes in Japan (adapted from [KIT 09])

7.5. Conclusions

This chapter illustrated a CCE study that identified three categories and 18 subcategories of information that drivers evaluated as necessary for safe and enjoyable driving. The mental processes for accomplishing train–station navigation tasks are slower than those for the tasks to follow the directions of a car navigation system (or a “human” car navigation system in this chapter), which is the example we illustrated at the beginning as shown in Figure I.2. The detailed workings of System 1 and System 2 would be different. However, since both share the time-critical features of interactions between human beings (passengers or drivers) and the environments, the case studies suggest that more suitable interactions design will be possible for those with specific cognitive characteristics for performing the tasks the interactions design should support satisfactorily. For example, for those suffering from information gathering tasks, or attention tasks, the critical indications, which are signboards in the case of train station and navigation directions in the case of car navigation systems, have to be placed where they expect to find them.

Information provided by a navigator is narrative in nature. Therefore, it should affect driver’s System 2 before mode if it is given for the future driving behavior, or System 2 after mode if it is given for the past driving behavior. Whichever mode the navigator’s information has influence on, the driver’s memory should have been altered in combination with the associated processes carried out in System 1 before and/or after modes. The particular driving episodes associated with the navigator’s information should have been processed consciously in System 2 before and/or after modes and therefore the memory trace created for the episode becomes consciously accessible when necessary in the future. We showed the role of conscious processing of an event while memorization for recalling it in the future consistent with the above-mentioned description through another CCE study where our participants were asked to watch short films at a theater [KIT 14b].

A practical implication found in this chapter is obtained by combining informational needs of drivers for safe and enjoyable driving as derived by the method described in this chapter with technological solutions that will make this happen. This information is very useful to consider how to design human–machine interface that should comfort to information processing processes of a driver as simulated by MHP/RT, which treats the cognitive functions that should work while driving, including working memory, attention, and planning that determine “how”, and memory resonance that determines “what” as an integral unified autonomous system. For example, for a driver who does not know the area and is not good at merging and changing lanes, it would be useful for him/her to be given such instruction as “change to the left lane now because there is a merge from the right shortly”. This is technologically feasible if there exists a system that can judge whether the following conditions are met:

  1. 1) the car is driving on the right lane;
  2. 2) there is a merge on the route;
  3. 3) it is possible to change to the left lane considering the current traffic condition;
  4. 4) the system can know the driver’s attributes as described above.

In the past, little serious effort has been made on matching drivers’ informational needs with technological considerations. The informational needs this chapter has identified, however, are not comprehensive. More studies will be necessary to investigate drivers’ informational needs covering all driving situations.