Abstract
Introduction
Physical fatigue while driving a car is partly related to perceptual-motor adjustments needed to perform the driving task such as using the pedals, and holding and operating the steering wheel [1]. Grujicic et al. [2], moreover, defined three factors related to physical driver fatigue: the maximum level of muscle activity and the number of muscles activated, shear force and spinal force. Thus, effects of sitting are also related to driving fatigue. The factors defined by Grujicic et al. [2] correspond to the three sitting characteristics that influence discomfort defined by De Looze et al. [3] based on a literature review: pressure distribution, muscle activity and lumbar curvature. Furthermore, Le et al. [4] found that blood oxygenation in leg muscles correlated to discomfort when seated, in general.
Consequently, prolonged static sitting is an aspect of driving which causes physical fatigue and, moreover, restricted postures also lead to a higher risk of musculoskeletal complaints [5]. Protracted sitting causes discomfort and, in general, it is encouraged to periodically engage in non-sedentary activities [6]. However, in the context of driving a car it is not possible to move vigorously without aborting the driving task. Ebe and Griffin [7] found that discomfort is primarily influenced by the seat characteristics and that vibration can be ruled out as an influence of physical fatigue in modern cars.
Several efforts have been made to develop ergonomically-sound seatsby attempting to define an ideal pressure distribution (i.e. Zenk [8]) or to support an ideal spine curve with a lumbar support (i.e. De Carvalho and Callaghan [9]). This is probably not sufficient to prevent discomfort; movement or frequent change between (healthy and stable) body postures is necessary to improve seating comfort [10]. Graf et al. [11] also supported natural movements (within an acceptable range) are desirable. Moreover, a study of Fujimaki and Noro [12] showed that natural movements occur in order to decrease discomfort caused by prolonged sitting. Although several studies examined the effects of body movements on discomfort, there seems to be no consensus on the frequency or range of movement. Vergara and Page [13] stated that a macro-movement is a distinctive change of posture every 5 to 6 minutes. Callaghan and McGill [14] defined dynamic sitting as variation of at least 3 positions over 2 hours. Several researchers investigated the effect of passive micro-movements.
Van Deursen [15] found that passive rotation of a pig cadaver spine resulted in an immediate increase of disc height at a rotation of less than 2°. However, Beach et al. [6] tested a continuous passive motion device in an office chair, which resulted in no difference in EMG or locally perceived discomfort. Passive rotation in an office chair compared with no passive micro-movement for the same office tasks, resulted in significantly shorter spinal length in the static chair [16]. A long term study on the effects of a similar passive motion device on low back pain found no advantage of the device [17]. Therefore, local passive micro-movements were not sufficient.
In order to offer dynamic sitting, office chairs have been developed with certain swinging mechanisms, resulting mostly in a greater variation of the inclination angles of the seat [18]. Muscle activity has been reported as an effect of sitting in this kind of chair [19]. Groenesteijn et al. [20] also found that such a swing-system chair is related to positive comfort evaluations in the context of computer work but is also very restricting.
The aforementioned studies indicated that postural change consisting of the variation of seating angles contributes to comfort rather than the application of local micro-movements. The concept here is to offer posture variation within the limitations of driving. The goal of this passive posture variation was to enable variation in pressure distribution and muscle activity to counter the physical fatigue from static seating. Moreover, this could result in more perceived comfort for car drivers. The first demand of passive posture variation is to offer variations in which the body can sense the movement but the difference in configuration could result in discomfort. Helander et al. [21] found noticeable differences in seat height, seat pan angle and backrest angle of office chairs and also the sensitivity of postural change resulted in response to body tissues and joints. However, allowing for posture variation in a car is particularly challenging because the driving task is a restraining posture and the driver cannot be distracted. Studies in office environments have already shown that the task restrains posture more than the actual seat [20, 22]. Therefore, the second requirement of passive posture variation should allow for various aspects of the driving task such as operating the steering wheel, pedals and vision in the mirrors. Thus, this study evaluated how much postural variation is allowed by the driving task and how sensitive is the human body is to these changes.
Consequently, this paper aims to answer the following research questions:
Can the body sense the difference in the succeeding steps of the seat movement? What seat movements do not affect vehicle operation (i.e. steering, using pedals, and mirrors)?
For this purpose, two experiments were carried out. The first experiment evaluated if and when seat configuration changes were noticeable to the human body. The second experiment evaluated the influence of the varying inclination of the backrest and the seat pan on typical aspects of the driving task in order to determine a possible range of motion.
Experiment 1: Noticeable differences in the seat configuration
In order to learn whether the difference in seat configuration is perceived by the human body, an adaptation of a method by Helander et al. [21] was developed. It was used it to find differences in office seat configuration which define how much adjustment the seat should allow. The hypothesis is that if the human body cannot perceive the differences, the variation in seat configuration will not be perceived as annoying or disturbing. Given that the seat movement was run in succeeding steps, the difference between those steps was evaluated.
Method
Twenty-participants (17 male, 3 female, 169–197 cm, aged 19–25, all students of Delft University of Technology) evaluated different seat configurations in a laboratory. For this experiment, a mock-up was built representing the ideal seat configuration according to Harrison et al. [23]. The mock-up allowed variations of both the inclination of the backrest and the seat pan. The position of the backrest were –6°, –3°, –1.5°, 0°, 1.5°, 3°, 6° and the positions of the seat pan were –4°, –2°, –1°, 0°, 1°, 2°, 4°, where 0° represents the seat angles according to Harrison et al. [23]. Figure 1 shows how the directions of seat adjustment were assigned. These same configurations were applied in experiment 2, except for the most extreme angles. The extreme angles were added to make sure that the participants would not get confused in case they did not notice any differences. The increment size of 1.5° for the backrest and 1° for the seat pan was based on the study of Helander et al. [21]. The mock-up was covered with a sheet.
For evaluation, the participant sat down and told the researcher their mark based on the scale presented in front of them accompanied with a schematic representation of the chair indicating the rating directions. Using the 11-point Likert scale ranging from “too flat” to “too steep” the participant evaluated the inclination of each configuration of the backrest or the seat pan. Only one of two positions was altered for evaluation, the other remained at 0°. Next, the participant stood up (but was not allowed to turn around) and the researcher adjusted the configuration. This evaluation was repeated for each of the seven positions of the backrest and the seat pan. The order of positions was varied systematically. Ten participants evaluated first the backrest and then the seat pan and repeated the action twice, the other group started with the seat pan.
The internal consistency of evaluations was measured with Cronbach’s alpha to assess if the same configuration resulted into similar scores among the subjects. Since the data were non-parametric, a Friedman test was conducted to assess whether the differences in seat configuration lead to significant differences in the evaluation.
Results
Table 1 shows the evaluation of each configuration. The table shows that +3° is the best-rated configuration for the backrest and that +1° is the best-rated configuration for the seat pan. Moreover, the most extreme inclinations are rated the worst for both the backrest and the seat pan and the ratings approach the best configuration in a logical order. The consistency of the ratings is questionable (0.7 > α> 0.6) for one backrest configuration (+3°) and one seat pan configuration (–4°), poor (0.6 > α> 0.5) for three backrest configurations (–1.5°, +1.5°, +6°), and even unacceptable (0.5 > α) for two backrest configurations (0°, –3°) and two seat pan configurations (+1°, +2°).
There was a statistically significant difference in subjective evaluation of the configuration for both the backrest (p = 0.000) and the seat pan (p = 0.000). Post-hoc analysis with a Wilcoxon signed-rank test was conducted with a Bonferroni correction applied, resulting in a significant level set at p < 0.008. Table 2 shows the results of this statistical analysis. For the backrest, only the two most extreme configurations (–6° and +6°) lead to a significant difference in evaluation. For the seat pan, the two most extreme configurations (–4° and +4°) were evaluated significantly different; however the difference in evaluation between 0° and +1° was also significant.
Experiment 2: Variation of seat configuration related to the driving task
It was necessary to learn whether movements of the backrest and seat pan were applicable within the limits of the driving task. The purpose of this experiment was to define a possible range of motion for a car seat with passive posture variation which still allowed performance of the driving task. Looking into the mirrors and operating the steering wheel and pedals were considered typical aspects of the driving task.
Method
Twenty-two subjects (14 male, 8 female, 161–186 cm, aged 23–46, all interns or employees working at the R&D centre of BMW) participated in the experiment. They were asked to sit in a full-size model of a car (four doors sedan) with a mechanically adjustable steering wheel, rear view mirror, production series pedals, and an electronic adjustable seat (seat height, seat position, seat pan angle, back rest angle, adjustable lumbar support and head rest position).
The participants were asked to adjust the seat, steering wheel and rear view mirror as they would when actually driving. Assuming this would be the best possible configuration for the participants in this particular model, the participants then rated visibility in the rear view mirror, accessibility of the pedals and position of the steering wheel. For this purpose a six-point rating scale was used (very good (3), good (2), rather good (1), rather bad (0), bad (–1), very bad (–2)).
Next, a series of adjustments were executed by the researcher while the participant remained seated. Each subject was asked to rate each new configuration on the three topics as compared with the starting configuration. After each adjustment, the seat was returned to the starting position. Thus, they could compare each seat configuration to their starting position. The adjustments (Fig. 1) increased the inclination of the backrest of +1.5° and +3° (backwards), decreased the inclination of the backrest of –1.5° and –3° (forwards), increased the inclination of the seat pan of +1° and 2°, and decreased the inclination of the seat pan of –1° and –2°. The adjustments were carried out separately. The type of seat used for the study adjusted the angles related to the H-point. For some participants, it was not possible to conduct all seat pan adjustments because the starting position was at the end of the range of the adjustment system. The seat angles were measured with a digital spirit level attached to the seat’s frame. Since the data were non-parametric and ordinal, the Wilcoxon Signed Rank Test was applied for statistical evaluation of significant differences (p = 0.05) between the evaluation for the different seat configurations.
Results
Figures 2 and 3 showed the average rating of the 22 subjects for the different backrest and seat pan adjustments on the following points:
View in the rear view mirror. Position of the steering wheel. Accessibility of the pedals.
The vertical axis represented the six rating levels and the horizontal axis shows the different positions of the backrest or seat pan (P1 was the starting position established by the participant). The six rating levels were very good (3), good (2), rather good (1), rather bad (0), bad (–1) and very bad (–2). In the starting position, the position of the steering wheel (2.00±0.976) and the accessibility of the pedals (2.14±0.640) is rated as good. The view in the rear view mirror, however, was rated neutral (1.05±1.174).
When adjusting the backrest, the rating of the accessibility of the pedals remained “good” for the positions –1.5° (1.95±0.0.844), +1.5° (2.05± 0.785) and +3° (2.05±0.785). For –3° in relation to the starting position the rating shifted to “rather good” (1.23±0.973). For adjustment of the seat pan, the rating of accessibility of the pedal remained “good” for all adjustments (–2°= 1.91± 0.831, –1°= 2.00±0.0.784, +1°= 1.80±0.894, +2°= 1.70±1.174). The position of the steering wheel was still rated as “good” for the backrest adjustment of +1.5° (1.68±0.1.041), but shifted to “rather good” for the adjustments of +3° (1.23±1.020) and –1.5° (1.23±1.193). The adjustment at –3°, for the steering wheel position was rated as neither good nor bad (0.36±1.364). The rating of the steering wheel position remained “good” for all seat pan adjustments (–2°= 1.91±1.136, –1°= 2.07±0.0.997, +1°= 1.90±0.995, +2°= 1.75±1.020).
While adjusting the backrest angle, the rating of the view in the rear view mirror shifted to neither good nor bad for +1.5° (0.18±1.097) and –1.5° (0.27±1.453). The adjustments of +3° (0.05±1.327) and –3° (–0.05±1.327), shifted the ratings to rather bad. The rating of the view remained “rather good” when adjusting the seat pan for +1° (1.00±1.183) and –1° (1.00±1.468). When adjusting the seat pan for +2° (0.65±1.309) or –2° (0.45±1.635), however, the rating shifted to neither bad nor good.
Tables 3 and 4 show the results of the Wilcoxon Signed Rank test conducted to determine if the rating difference between the various adjustments was significant. When adjusting the backrest (Table 3), there was a significant difference in the rating of the view in the rear view mirror for +1.5° (p = 0.005) and –1.5° (p = 0.006). During further adjustment of the backrest, this rating did not change significantly from +1.5° to +3° (p = 0.166) or –1.5° to –3° (p = 0.088). Furthermore, backrest adjustments of +1.5° (p = 0.066) and from +1.5° to +3° (p = 0.067) do not change the rating of the steering wheel position significantly. However, the rating for +3° differed significantly from the starting position (p = 0.014). For the adjustments of –1.5° (p = 0.005) and –1.5° to –3° (p = 0.004), there was a significant difference in the rating of the steering wheel position. The first steps of adjusting the backrest (+1.5° and –1.5°) had no significant influence on the rating of the pedal position (p = 0.157 and p = 0.102). The next steps, however, did significantly change the rating (p = 0.000 for +1.5° to +3° and p = 0.002 for –1.5° to –3°).
The seat pan adjustment (Table 4) of +1° did not have a significant influence for rating of the view in the rear view mirror. This rating, however, differs significantly for the next adjustment from +1° to +2° (p = 0.034). Adjusting the seat pan in the other direction (–1°) also resulted in a significant differences in rating (p = 0.046), although the next adjustment from –1° to –2° did not elicit further significant change (p = 0.059). Adjusting the seat pan angle did not result in significant differences in the rating of the position of steering wheel and pedals for all steps.
Discussion
The results of experiment 1 showed that the configuration of backrest and seat pan rated best but did not correspond to the ideal seat angles according to Harrison et al. [23]. This difference (backrest +3°, seat pan +1°) could be explained by the lack of context (vehicle interior, driving task). Moreover, the fact that the consistency of the ratings was questionable to unacceptable, for nine of fourteen configurations showed that the differences in configuration where so small that the participants did not evaluate the same angle twice in a similar way. The lack of significant differences between the configurations appeared to confirm this. However, the difference between a seat pan angle of 0° and a seat pan angle of 1° was an exception, since this difference was statistically significant. Previous research also showed that, depending on the vehicle, there was an error in noticing the repeatability of the seat position for the backrest of 1.7°–2.9° and for the seat pan of 1.1°–1.7° [24].
The driving task was the critical restraint for posture variation. Experiment 2 shows that a seat pan adjustment of 1° upwards had no significant influence on the rating of the interior configuration for this seat and could therefore be applied to passive micro-movements. The seat pan adjustment of –1° and the backrest adjustment of +1.5° did not significantly influence the ratings on steering wheel and pedal position. This demonstrated that the most critical feature was the rear view mirror. Changing the backrest position for –1.5° also influenced the rating of the steering wheel position significantly. The rating of the rear view mirror did shift to rather bad, but the rear view mirror was also rated worst than both steering wheel and pedal position in the starting configuration (the rear view mirror seemed too small for this type of vehicle). This implies that if the rear view mirror was rated as “good” for the starting position, the seat pan adjustment of –1° and the backrest position of +1.5° were also acceptable for visibility. Future developments in the direction of a screen instead of a rear view mirror perhaps would make this factor no longer a constraint.
A limitation of this experiment was that adjustments were executed sequentially with the participants seated and it is possible that this made the difference more noticeable. However, this method was chosen in order to allow participants to evaluate the differences from the starting configuration. The participants were very conscious of what they thought to be the logical implications of the adjustments and this may have been a cause for more sensitivity. The question remains that whether the ratings would have differed so strongly if participants got out of the car between the various configurations and were unaware of the adjustments. Experiment 1 shows that consistency of ratings was not good. The reactions of the participants on the researcher adjusting the seat while they were seated, however, also show the importance of low adjustment velocity when applying this range of motion for passive micro-movements. The results of these experiments could have been influenced by the seat type. Depending on the stiffness of the cushions, the seat compensated for some of the movement. Therefore, it is possible that a different seat would result in a slightly different outcome. Furthermore, this study was executed in a lab and it is not known how these posture variations would influence the ratings when actually driving a car. Another limitation of the study was that the view on the instruments panel and side mirrors were not considered and left out. However one could assume that the effects of these entities are similar to the rear view mirror.
The goal of this study was to derive recommendations for a possible range of motion. For passive micro-movements, a variation in seat pan angle between –1° and +1° and a variation of backrest angle between 0° and +1.5° seemed plausible (Fig. 4). In the future, it would be of interest to learn the effect of such a range of motion for passive micro-movements on subjective and objective measures of comfort. Moreover, an experiment during actual driving would verify whether the variation really was not perceived as annoying or disturbing.
Conclusion
The difference in between the various seat configurations appeared to be so small that the participants were not able to evaluating the same angle twice in a similar way. Moreover, only two most extreme configurations (–6° and +6°) of the backrest lead to a significant difference in evaluation. For the seat pan, the two most extreme configurations (–4° and +4°) were also evaluated significantly different, and the difference in evaluation between 0° and +1° was significant. The driving task appeared to be the critical constraint for posture variation. Adjusting the seat pan angle for +1° had no significant influence on the rating of the interior configuration (visibility in the rear view mirror, accessibility of the pedals and position of the steering wheel) by the participants. Adjusting the seat pan angle for –1° or the backrest angle for +1.5° has no significant influence on the ratings of steering wheel and pedal position. When a better starting configuration for the rear view mirror is provided, a range of motion is recommended with a variation of seat pan angle between –1° and +1° and a variation of backrest angle between 0° and +1.5°. If decreased adjustment velocity results in less sensitivity for the users, variation of backrest angle between –1.5° and +1.5° was considered the starting configuration of the rear view mirror. Although the rating difference might be significant, the shift from “good” to “rather good” can still imply an acceptable interior configuration. Further research is needed to evaluate the effect of passive posture variation on subjective and objective comfort measures.
Conflict of interest
The authors have no conflict of interest to report.