How drivers sit in cars
Post on 15-Jun-2016
Embed Size (px)
Accid. Anal. and Prev., Vol. 27, No. 6, 777-783, 1995 pp. Couvrieht 0 1995 Elsevier Science Ltd Prkzin the USA. All rights reserved
oool-4575/95 $9.50 + 0.00
HOW DRIVERS SIT IN CARS
S. PARKIN*, G. M. MACKAY and A. COOPER
Birmingham Accident Research Centre, University of Birmingham, P.0 Box 363, Birmingham B15 2TT, U.K.
(Received 26 August 1993: Accepted 1 January 1994)
Abstract-This paper presents results from a study to measure the separation of the drivers head and shoulder to various internal features of the car. Drivers were filmed whilst driving in general traffic flow, hence were unaware that they were involved in a study. The results show that certain sub-groups of the driver population are likely to be more at risk for certain impact types. Small females are considerably closer to the steering wheel than the rest of the population, and therefore prone to head strikes in frontal impacts. Large males are likely to interact with the cant rail and B-pillar in side impacts.
Keywords-Drivers, Sitting position, Cars, Video, Hybrid III, Dummy, FMVSS 208, FMVSS 214, Restraint, Airbag, Head restraint
Much design work related to crash performance is predicted on the initial sitting positions of current crash test dummies. In-depth crash reconstruction aimed at evaluating seat belts, steering wheel design and airbag performance is similarly based on assump- tions about sitting position and the posture of drivers. Several studies have attempted to quantify how driv- ers actually sit in cars, but these have generally been laboratory based, and therefore not necessarily repre- sentative of the real-world driver population under real-world driving conditions. The data used to posi- tion dummies in the current crash tests came from a NHTSA sponsored study at UMTRI (Robbins et al. 1983) in which the subjects were seated in a standard- ized driving posture, with the seat back angle fixed. The subjects were asked to sit upright, with their back pressed against the seat. In a study to determine the drivers eye position, Meldrum (1966) measured subjects seated in actual vehicles, but under simulated driving conditions, placing the vehicle in front of a large street scene mural. The data from this work led to the drivers field of vision standard SAE 5941.
Other studies have attempted to measure drivers during, or after, driving. Schneider et al. (1979) com- pared driver anthropometry in non-driving and after- driving conditions, and found little overall differences in the two situations. This study, however, took no account of seated posture. A study at Jaguar Cars by Wankling ( 1991) measured the drivers eye position under actual driving conditions, using an electromag-
*Author for correspondence.
netic sensor, but no firm conclusions were reached due to the small sample size. In a recent study, Schneider et al. (1991) compared measured driver eye position to SAE 5941, from a sample of 50 subjects and 6 vehicles after driving on a short test track.
This paper will present results from drivers, driving their own cars, under actual driving condi- tions, without the subjects being aware that they are being involved in a study.
Drivers were filmed using a video camera equ- ipped with a high speed shutter as they passed a white screen. Three separate locations were used, with the camera being out of sight of the drivers. The camera was pointed at right angles to the traffic flow, facing the screen, as shown in Fig. 1.
Hence, the drivers were silhouetted against the
rl Video Camera Fig. 1. Unobtrusive filming of the drivers.
778 S. PARKIN et al.
background. The camera was able to capture the image of the vehicle and driver, even when the vehicle was travelling at around 30 mph. The camera height was set at mid side window height for an average vehicle, and filming took place at between 15 and 20 m from the subjects. The angle from the camera to mid side window and top of the side window was therefore very small, and so parallax was not consid- ered to unduly affect results. The video was played back and measurements taken from a television moni- tor. Figure 2 shows which measurements were taken: nasion (junction of the forehead and nose) to the steering wheel top (A-B), and to the top of the side window (A-D); the centre of the shoulder to the centre of the B-pillar (I-J); and front or back of the head (whichever was in view) to the centre of the head restraint (A or G-H). The B-pillar height was measured and used as a scaling factor for all the measurements when compared to the known height of the B-pillar. A correction factor was also applied, as explained later in the paper.
After the measurements were scaled and cor- rected, they were converted to the following dimen- sions: nasion to steering wheel centre (horizontal, vertical and direct; A-C); top of head to top of side window (vertical; E-F); back of head to centre of head restraint (horizontal and vertical; G-H); and centre of shoulder to centre of B-pillar (horizontal, I-J). The measurements to the top of the side window and the centre of the head restraint required anthro- pometric data to shift the originally measured point on the head to the required point on the head (A-E, and A-G).
In order to calibrate the system, five subjects of varying age, sex and stature were filmed in two different car types from a number of filming distances. The drivers head was fixed in a stable position using the cars head restraint. The actual forehead to steer-
Fig. 2. Measurements taken of the driver to the vehicle.
ing wheel separation was measured and then com- pared with measurements taken from five filming distances, both static and dynamic. In the dynamic condition, the car was driven at between 15 and 20 mph. Results indicated that the on-screen measure- ments were all between 1 and 11% greater than the actual measurements. A correction factor of -6% was therefore applied to all subsequent measurements making them accurate to within f5%.
The cumulative frequency plots shown later in the paper have been analysed using the probability density function of normal distributions, and the standard deviations at various percentile levels have been calculated. The calaculated values for twice the standard deviation of a particular percentile value is given underneath each figure (rounded to the nearest mm), with the percentile value it refers to also shown. A line may be drawn horizontally on the cumulative frequency curve at the particular percen- tile value such that the line is centred on the actual value and extends twice the standard deviation in each direction. The line represents a 95% probability that the percentile quoted lies within its bounds. In general, the line would be so small (usually of the order of lo-20 mm) that it has not been practical to plot it on the curves.
Nineteen car models were chosen based on the most popular models from 7 manufacturers. Three of the manufacturers had most of their range in the sample, from the smallest to the largest model. The car sample was thus thought to reasonably represent the general car population in terms of size and weight. As each of the sample vehicles passed the camera in the general traffic flow, the measurements were
Table 1. Driver population by sex
Male 742 Female 237 Not known 21 Total 1000
Table 2. Driver population by age
Young 315 Middle 564 Old 78 Not known 43 Total 1000
32.9 58.9 8.2
How drivers sit in cars 179
Table 3. Vehicle population by make and model
Make and Model Number
Citroen BX 26 Ford Fiesta MK3 49 Ford Escort MK3 94 Ford Escort MK4 58 Ford Sierra 148 Ford Granada 21 Peugeot205 47 Peugeot 405 33 Rover Metro 38 Rover 200 69 Rover Montego 96 Rover 800 24 Vauxhall Nova 41 Vauxhall Astra 81 Vauxhall Cavalier 63 Vauxhall Carlton 17 Volvo 760 19 Volkswagen Polo 39 Volkswagen Golf 31
Table 4. Vehicle population by wheelbase and size
2.05-2.40 2.41-2.58 2.599280 Total
Crash 3 size
1 2 3
218 432 350
recorded. In this way the driver population generated itself, and was completely random. One thousand readings were taken. Tables l-4 show how the driver and vehicle population was comprised. In the age category Young means the driver looked to be below 35 years old; Middle, the driver appeared to be aged 35-55; Old, the driver looked to be aged over 55.
The majority of the driver sample was male (75%) and aged between 35 and 55 (60%).
The following graphs (Figs 3-5) are the plotted results for each of the measurements taken. For each graph the sample is plotted as a whole, and also split by sex. Each sub-group has the following percentiles plotted: 1, 5, 25, 50, 75, 95, 99. It should be noted that each percentile plotted is the percentile within the measurements and does not necessarily represent the overall population percentile. Hence, when the paper quotes 50%ile male it means the 50%ile measurement of the male sample, not the measure- ment that occurs with the 50%ile male.
Nasion to steering wheel centre Figure 5 shows that females are considerably
closer to the steering wheel than males.
--o- ALL (n=lOOO) __*_ Male (rl=741)
._-__t-._ Female (n&237)
150 200 250 300 350 400 450 500 550 6 Horizontal distance (mm)
Fig. 3. Nasion to steering wheel centre (horizontal).
Twice standard deviation