Relationship between Senior Drivers’ Cognitive and Physical Functions and Urban Driving Behaviors during Behind-the-Wheel Driving Education Sessions with Driving Instructors

Recently, traffic accidents caused by elderly drivers have been increasing. It is thought that the reasons of accidents are functions such as decline of eyesight, cognition and physical strength caused by aging. In order to assist safe driving for elderly drivers, it is necessary to sufficiently understand any possible relationship of various senior drivers’ cognition, physical strength and driving behavior. In this paper relationship of elderly driver’s cognition, physical strength and driving behaviors were analyzed using the result of driving instructions on urban road by driving school instructors.

• Human Engineering
• Elderly Person
• Driver Characteristics
• Driving Instruction

1. Introduction

In Japan, the growing elderly population has led to an increased proportion of elderly drivers; accordingly, the number of traffic accidents caused by elderly drivers is also increasing [1] . To address this matter, this study was conducted as part of the Nagoya Center of Innovation (COI) program promoted by the Ministry of education, culture, sports, science and technology and the Japan Science and Technology Agency (JST) in an attempt to safely delay the cessation of driving for the average elderly driver [2] .

Delaying the cessation of driving requires driver education and the development of support systems for the elderly that consider the relationship between aspects of driving, such as decision-making and physical control while driving, and human characteristics such as physical attributes and cognitive function [3] . According to data on traffic accidents, the second most frequent collision type, after rear-end collisions, is bumping at intersections (crossing collisions), and the data show that seniors, in particular, are often involved in crossing collisions [4] . Additionally, elderly drivers have been shown to exhibit delayed response during emergencies at intersections [5] and insufficiently check both sides at stop sign intersections [6] . In order to understand these challenges, our experiment provided behind-the-wheel training to elderly drivers on city streets that included intersections with stop signs, where it is easy for crossing collisions to occur. This behind-the-wheel training was given by a driving instructor from a driving school who was present in the car with each elderly driver [7] . In our previous report, the content of the instructions given by the driving instructors was analyzed and a proposal was made for a system to assist with driving instructions for the elderly [8] . This report investigates the relationship between the frequencies of instructions given to senior drivers at stop sign intersections and driving behaviors, physical strength, and cognitive function; it also discusses a proposed method for supporting safe driving by the elderly.

2. Experimentation Methods

2.1. Subjects

The subjects consisted of a total of 24 elderly people (Id1 ~ Id24) comprising 12 males and 12 females. One of four driving instructors (Is1 ~ Is4) rode with each subject and provided instructions from the passenger seat. Table 1 shows the combinations of subjects and driving instructors and the conditions under which the session was conducted.

2.2. Measurement Methods

The elderly drivers were asked to follow the circular driving route shown in Figure 1. This course (intersections 5 → 6 → 9 → 1 → 4 → 5) is located on residential streets near Nagoya University, and the elderly drivers completed it in 17 to 30 minutes. This circular course included 10 intersections: three were stop sign intersections where a left turn was made, two were stop sign intersections where the driver proceeded straight, one was an intersection with a convex mirror where a left turn was made without stopping, and four were intersections where the driver proceeded straight without stopping (one of these had a flashing yellow signal). Many of these stop sign intersections had poor visibility, and there were two places where the route passed over a pedestrian crossing: one intersection where a left turn was made without stopping and one intersection where the driver proceeded straight without stopping. Half of the driving route was on

Table 1

Details of driving instructors and subjects

Is	Is: Age (Instruct years)	Id: No	Id: Age	Gender	Start time	Lesson duration	Weather
Is 1	42 (18)	1	57	Female	9:29:39	0:22:49	Fine
		2	54	Female	10:37:30	0:24:01	Fine
		3	74	Female	10:34:03	0:24:54	Cloudy
		4	75	Female	9:28:57	0:17:36	Cloudy
		5	75	Male	9:30:35	0:22:48	Rain
		6	70	Male	10:32:48	0:23:08	Rain
		7	70	Male	9:24:08	0:24:32	Fine/snow
		8	77	Male	10:33:49	0:20:49	Fine/snow
Is 2	46 (19)	9	67	Female	14:07:33	0:17:08	Fine
		10	76	Female	15:07:27	0:19:51	Fine
		11	70	Male	13:58:59	0:32:47	Cloudy
		12	68	Female	15:16:30	0:31:54	Cloudy
		13	54	Female	13:55:36	0:28:09	Rain
		14	65	Female	15:05:33	0:26:34	Rain
Is 3	48 (23)	15	66	Male	9:19:19	0:23:18	Fine
		16	59	Male	10:29:00	0:21:58	Fine
		17	68	Male	9:15:40	0:22:50	Fine
		18	64	Female	10:25:28	0:24:49	Fine
Is 4	44 (18)	19	79	Female	14:13:47	0:20:05	Fine/snow
		20	73	Female	15:16:18	0:27:31	Fine/snow
		21	52	Male	14:01:42	0:30:01	Fine
		22	77	Male	15:11:44	0:29:52	Fine
		23	50	Male	13:45:07	0:32:51	Fine
		24	55	Male	14:55:44	0:27:27	Fine

one-way roads. The roads on the course, including both one-way roads and two- way roads without center lines, were narrow enough to require driving slowly and intentionally making room to avoid cars in the opposite lane, parked cars, pedestrians, and bicycles.

As the subjects were not driving their own vehicles, they were given time to practice driving the vehicle provided while the driving instructors accompanied them in the passenger seat. After they had mastered the vehicle and the driving instructor had given them instructions on driving, including navigation and the operation of the auxiliary brake, they began their experimental drive on the public roads [7] . Prior to this, cognitive function tests such as the Mini-Mental State Examination (MMSE), Trail-Making Test (TMT) Parts A & B, and Useful Field of View (UFOV) test [a visual information processing speed task from the Driving Health Inventory (DHI)] had been conducted on all subjects. In addition, the subjects’ suitability for driving was examined using the National Police Agency’s Driving Aptitude Testing Device (Takei Scientific Instruments Co., Ltd.). The relationship between these test results and the frequency of instructions that the instructors provided to each driver while on the road was investigated [2] . Driving data was recorded with a BU-DRHD421 driving recorder (Yupiteru) that was equipped with a GPS (Global Positioning System) system and an acceleration sensor. Footage from a camera facing out from the front of the vehicle and that from a camera aimed at the driver were continuously recorded. Figure 2 shows an example of the recorded data.

Figure 1

Driving route of experiment

Figure 2

Example of driving-recorder viewer

After they were provided with an explanation of the content of the experiment, informed consent was obtained from all subjects and instructors before the experiment. This experiment passed an ethical review performed by Nagoya University’s Institute of Innovation for Future Society.

2.2. Analytical Method

After each experimental drive, the content of the instructions given by the instructors was written down from the recording voice of driving-recorder (Figure 2) and analyzed as follows. On the basis of the situation in which each instruction was given, including the location and driving behavior, the timing of each instruction was classified into one of the following five categories: in advance, immediately before, during a maneuver, immediately after, or as a review. The intention of each instruction was then sorted into one of the following five categories on the basis of its content: information, warning, instruction, navigation, and non-instruction for passing through an intersection. The content of each driving instruction was classified into one of the following eight categories: completely stop; stop before the stop line; brake; check left and right; go slowly; keep left; start moving; and (how to) navigate. The frequency of advice in each content category was calculated. Table 2 shows examples of the content of the driving instructions that were extracted from the drive recorder.

Furthermore, the time at which each braking maneuver started () and ended () was identified from the vehicle acceleration data as time data points, as shown in Figure 3. Data from the front-view camera (Figure 4) was used to identify the moment at which the vehicle’s hood touched the center mark of an intersection as the time of intersection entry ().The speed calculated by GPS signal was used, and the acceleration was used the value of the acceleration sensor of the drive recorder. Because the speed by GPS signal had a time lag of one second or more at the sampling frequency (about 1 Hz) of the GPS signal, the acceleration did not coincide with the gradient of the speed-time graph in Figure 3. However, the speed was not calculated by the integrated value of acceleration, because the acceleration sensor signal included the drift of the signal and the road slope error. Therefore, the average acceleration and the deceleration distance are calculated by the speed change per time. Table 3 shows example of driving behavior calculations of the Figure 3 graph.

The velocity at the start of braking () and that at the end of braking () were calculated using speed. In addition, the following braking information was calculated: the maximum deceleration for a single application of the brake (), time of continuous braking (), average deceleration for a single application of the brake (), constant acceleration ratio of the maximum deceleration to the average deceleration as an indicator of the smoothness of braking (), braking distance required for deceleration (), and time margin for starting to brake (). The time spent checking left and right was calculated as the interval between the completion of braking and the time of intersection entrance

Table 2

Example of driving instruction analysis

No	Map No	Road situation	Driving	Time	Driving instruction comment	Instruction intention		Timing	Driving instruction
1	10 → 11	One-way traffic	Straight ahead way	9:29:20	We'll turn left at that red flashing signal.	Navigation		In advance	Navigating
2	11	One-way intersection	Straight ahead intersection	9:29:24	It’s slow. I have a bad perspective. Slow down.	Instruction		Immediately before	Slowing down
3	11	One-way intersection	Straight ahead intersection	9:29:28	How to accelerate slowly.	Instruction		Immediately after	Mild accelerating
4	12	One-way stop sign intersection	Straight ahead stop sign	9:29:31	Stop it. Stop so as not to cross the stop line. Yes, brake, brake, and firmly brake.	Caution	Instruction	Immediately before	Braking	Stopping	Stopping position
5	12	One-way stop sign intersection	Straight ahead stop sign	9:29:39	Also, checking while looking at it.	Instruction		During maneuver	Left and right checking
6	12 → 5	One-way traffic	Straight ahead way	9:29:43	I will make a left turn on that stop. Turn left here.	Navigation		Immediately before	Navigating
7	5	One-way red flashing signal intersection	Left turn stop sign	9:29:46	Firm brake. Such crossing car may come.	Instruction	Caution	Immediately before	Braking
8	5	One-way red flashing signal intersection	Left turn stop sign	9:29:53	So red flashing signal is stop. It is same meaning as a stop sign.	Information	Caution	Immediately before	Stopping
9	5	One-way red flashing signal intersection	Left turn stop sign	9:29:58	Okay, we go ahead little by little. While turning steering wheel. Let's apply brake more.	Instruction	Instruction	During maneuver	Left and right checking	Slowing down	Steering
10	5	One-way red flashing signal intersection	Left turn stop sign	9:30:04	Because it is narrow, return steering wheel.	Instruction		During maneuver	Steering
11	5 → 6	Two-way traffic both side parked vehicles and a pedestrian	Avoid parked vehicles and a pedestrian	9:30:09	Let’s keep on braking. It's narrow. Brake, brake. More brakes, more brakes.	Instruction	Instruction	During maneuver	Braking	Slowing down

(), and the number of times that each driver checked left and right () was determined from the footage of the driver. The above performance indicators were used to investigate the relationship between drivers’ driving behaviors and the frequency of instructions by their instructors. The data on driving performance recorded by the drive recorder may have been imperfectly precise. Nevertheless, this method was deemed to be the best way to relatively analyze the behavior of the instructor and that of the driver.

Figure 3

Example of the driving-recorder data analysis

Figure 4

Intersection entering timing ( )

Table 3

Example of driving behavior calucration of the Figure 3 graph

Driving behavior	Symbol and Equation	Value
Braking start speed		24 km/h
Braking end speed		1 km/h
Braking start time		2.0sec
Braking end time		6.9sec
Braking time		4.9sec
Braking distance		16.3m
Time margin of braking		2.45 sec
Braking average deceleration		2.06 m/sec2
Ratio of max and average deceleration		1.51
Intersection entering time		9.8 sec
Left and right checking time		1.9 sec
Number of left and right checks		4 times

3. Experiment Results

3.1. The Content of Instructions Given by Instructors

Figure 5 shows the distribution of driving locations where instructors gave instructions (3117 instances) in all of the experimental drives by the 24 senior drivers. This study analyzed the instructions and driving behaviors for all instances at stop sign intersections (1004 instances), except in situations where things such as other vehicles had an interfering influence at the intersection.

Figure 6 shows the distribution of the intentions of instructions given at the stop sign intersections that were analyzed here. The driving instructions were carefully counted, and the drivers who received many driving instructions were considered to have poor driving performance. The relationship between instruction frequency on one side and cognitive attributes, physical attributes, and driving behaviors on the other was examined.

Figure 7 shows the distribution of the content of instructions that were given by the instructors, including advice on navigation, information, warnings, driving instructions, and driving interventions, at stop sign intersections. At stop sign intersections, the instructors frequently gave instructions to stop completely, stop before the stop line, and check left and right while moving forward slowly.

3.2. The Relationship between Driving Behaviors and the Frequency of Driving Instructions

Table 4 shows the driving instruction frequency and measured average data of driving behavior, cognitive function tests and driving aptitude tests of each subject. Based on this data, the correlation coefficient between each function was

Figure 5

Distribution of driving situations where instruction was given (Total: 3117)

Figure 6

Distribution of the intention of instructions given at stop sign intersections

Figure 7

Distribution of the types of driving maneuvers that warranted instructions at stop sign intersections

calculated using the Equation (1)-(5) and the relation of each function was studied. As there was a great deal of variation in human characteristics and human behaviors, it is very difficult to obtain a statistically significant confidence level for the correlation coefficient, requiring prohibitively numerous subjects and an

References

1. National Police Agency (2015) Statistical Traffic Accident Data. http://www.npa.go.jp/toukei/koutuu48/toukeie.htm

2. Aoki, H., Kanamori, H., Yamagishi, M., Tanaka, T., Takahashi, I., Yonekawa, T., Kawano, N., Ito, Y., Iwamoto, K., Ozaki, N., Terasaki, H., Obinata, G., Akamatsu, M., Sato, T., Oguri, K., Kawanaka, H. and Nakagawa, T. (2015) Study on Driver Characteristics for Delaying Driving Cessation (1) Database Construction of Older Drivers’ Human, Aging, Driving Characteristics (in Japanese with English summary). JSAE 20155208, 2015 JSAE Annual Spring Congress Proceedings, Yokohama, 20-22 May 2015, 1091-1094.

3. Yamagishi, M., Aoki, H., Tanaka, T., Takahashi, I., Kawano, N., Sato, T., Aakamatsu, M. and Kanamori, H. (2015) Study on Driver Characteristics for Delaying Driving Cessation (2) Correlation Analyses between Fundamental Characteristics of Older Drivers and Their Driving Behavior (in Japanese with English summary). JSAE 20155209, 2015 JSAE Annual Spring Congress Proceedings, Yokohama, 20-22 May 2015, 1095-1100.

4. Institute for Traffic Accident Research and Data Analysis (ITARDA) (2007) Accidents Involving Elderly Drivers. 2007-06, ITARDA Information, No. 68. http://www.itarda.or.jp/itardainfomation/info68.pdf

5. Yonekawa, T., Murano, T. and Iwazaki, K. (2014) Analysis of Accident When Turning Right at Intersection Using Driving Simulator. International Journal of Vehicle Safety, 7, 17-36. https://doi.org/10.1504/IJVS.2014.058239

6. Yonekawa, T., Sasaki, K. and Iwazaki, K. (2014) Driving Behavior Analysis of Elderly Drivers Passing through Stop Sign Intersection Using Driving Simulator. International Journal of Automotive Engineering, 5, 137-144.

7. Y. Yoshihara, E. Takeuchi, Y. Ninomiya (2015)Acquisition and Analysis of Driving Data of Expert and Elderlies in Residential Area toward Developing a Referential Driving Behavior Model Adapted for Elderlies (in Japanese with English summary). JSAE 20155241, 2015 JSAE Annual Spring Congress Proceedings, Yokohama, 20-22 May 2015, 1246-1251.

8. Yonekawa, T., Tanaka, T., Aoki, H., Yamagishi, M., Yoshihara, Y., Takeuchi, E., Takahashi, I., Ninomiya, Y. and Kanamori, H. (2015)Analysis of Instructor’s Teaching of Senior Drivers’ Safety Urban Driving Behavior (in Japanese with English summary). JSAE 20156054, 2015 JSAE Annual Autumn Congress Proceedings, Kitakyushu, 14-16 October 2015, 276-281.