About this Journal Submit a Manuscript Table of Contents
Advances in Mechanical Engineering
Volume 2013 (2013), Article ID 171479, 10 pages
Research Article

Study of Correlation between Driver Emergency Measures and Pedestrian Injury Based on Combined Driving Simulator and Computer Simulation

1State Key Laboratory of Automobile Safety and Energy, Tsinghua University, Beijing 100084, China
2Key Laboratory of Evidence Science of Ministry of Education, China University of Political Science and Law, Beijing 100088, China

Received 24 September 2013; Accepted 21 October 2013

Academic Editor: Fenyuan Wang

Copyright © 2013 Quan Yuan et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.


Driving simulator and computer simulation are combined to reconstruct the kinematic process of pedestrian-vehicle crash. Firstly, in-depth data from 158 accident cases in Beijing is analyzed. Additionally, the typical accident factors, collision characteristics, road status, and drivers’ emergency behavior are classified and integrated to establish virtual scenarios of pedestrian vehicle accidents. Moreover, volunteer drivers are tested on the driving simulator while data of vehicles’ real-time kinetic parameters from different emergency measure stages is recorded. Furthermore, the process of pedestrian-vehicle crash is modeled and simulated by accident analysis software PC-CRASH to obtain the data of pedestrian’s head centroid acceleration, and the HIC value is calculated as an injury indicator. Finally, the collected data is analyzed to find out the relation between drivers’ various emergency measures and pedestrians’ injury severity. The results show that steering with braking is the most effective measure among drivers’ various measures. This paper may provide some valuable suggestions on research of driving safety as well as intelligent transportation.

1. Introduction

As vulnerable road users, pedestrians are mostly at risk in urban areas due in part to the large amount of pedestrian and vehicle activity in urban areas, which will become more susceptible to traffic crashes, especially in developing countries such as India and China [1]. In various automobile accidents of China, pedestrian-vehicle accidents have large fatality. According to Traffic Administration Bureau of Chinese Public Security Ministry, pedestrians’ death caused by road traffic accidents in China accounted for 25.37% of the total in 2012 [2]. Therefore, there is great need to find out effective methods for the research and analysis on the safety protection of pedestrians.

The research of Mizuno and Kajzer indicated that the injuries of pedestrian were related to the human body height, impact speed, and frontal profiles of the vehicle [3]. Based on the Pedestrian Crash Data Study (PCDS) of National Highway Traffic Safety Administration (NHTSA), a demographic analysis and reconstruction of selected cases were carried out to determine if the precrash motion of the pedestrian and vehicle could somehow be linked to the injuries and vehicle damage [4]. The emergency measures adopted by drivers, for example, steering and braking, significantly influence the real-time kinetic parameters for vehicles, for example, braking deceleration, collision speed, and steering angle, which are related to pedestrians’ injury severity.

As an important research method for modern automotive engineering, driving simulators have become an increasingly widespread tool to understand and assess traffic safety [5]. One of special studies centers on the individual differences and driver state. By means of driving simulator, which has good test reproducibility in the analysis of the driver emergency response under complex working conditions, the different driver behavior can be tested in a safe environment. A driving simulator was developed to study a driver’s mental state in a traffic accident occurring between vehicles at an intersection [6]. The experimental results showed that the driving simulator could be used to study the relationship between driver's physiological data and the mental state. Driver performance in the moments during microsleep was studied by licensed drivers based on a high fidelity driving simulator [7]. Results indicated that driving performance deteriorates during microsleep episodes, which is important for the design and implementation of countermeasures such as drowsy driver detection systems. At Ford Motor Company, a fixed-base driving simulator study was conducted to analyze the driver workload effects of cell phone, music player, and text messaging tasks, which are the common risk factors related to crash [8]. As significant risk factors, alcohol and distraction were combined, and their synthetical effects on driving performance were examined successfully by a driving simulator [9]. Besides, driving simulator can be used in the product development, education, entertainment, and other research fields [1012].

The severity of an accident is largely depending on the driver’s behavior when traffic accidents occurred. The emergency measures taken by the driver when a car accident may possibly occur (emergent collision avoidance measures) directly affect the kinematic parameters of vehicles, for example, deceleration, speed, and front wheel’s steering angle of vehicle before collision. However, these parameters are directly related to the injury severity of vulnerable road users like pedestrians. There are two methods by which drivers may take to avoid collision: the first one is to steer the car (vehicle’s lateral movement) and the second one is to control the longitudinal movement like braking.

Many studies have discussed emergency measures to avoid accidents. Xuemei Chen’s study, which is about drivers’ braking behaviors under emergency, showed that the control of brake pedal under emergency can be divided into four stages [13]. Similarly, Kim et al. studied different emergency measures among different driving experiences, which discussed the factors of steering wheel angle and brake pedal pressure, and concluded that steering is a more advanced emergency decision-making method than braking [14]. However, if we have not the records for driving, it remains hard to take drivers’ emergency reaction into consideration in existing accident analysis.

In fact, as mentioned above, the collision avoidance methods of drivers have a significant relationship with accident severity; therefore, in our study, we will combine the methods of driving simulator and computer simulation to examine the relationship. Considering the fact that head injury is the most common fatal injury in vehicle accidents, our study will have a discussion on Head Injury Criterion (HIC) value of pedestrian.

2. Equipment and Method

2.1. Driving Simulator Equipment

The driving simulator system we used in this study is shown in Figure 1. This system consists of visual simulation systems, sound simulation system, touch simulation system, vehicle dynamics simulation systems, central control platform, data base and program base, and other components, which uses a combination of many advanced technologies like three-dimensional images technology, virtual reality (VR) to reconstruct the view of a real traffic environment and interact driver’s sensation of movement and operation to simulate a driver-vehicle-environment closed-loop system.

Figure 1: Driving simulator system.

In this study, we use the driving simulator to obtain the real drivers’ emergency measures to avoid the collisions with pedestrians in the virtual scenario.

2.2. Computer Simulation Software

PC-CRASH is software for accident reconstruction that DSD Corporation developed. Due to the advantage of easy modeling, good simulation, and completeness of function, it is now widely used to reconstruct and simulate accidents [1518]. In this study, integrated with the driving simulator, this software is utilized to simulate the car-pedestrian collision process, which usually has 5 stages as follows.(i)To select the simulation model from the database.(ii)To input the parameter information of vehicle.(iii)To input the parameter information of pedestrian.(iv)To confirm the contact position between vehicle and pedestrian.(v)To carry out the simulation of the collision process.

2.3. Research Method

Useful information is extracted from a large number of accident cases, to build typical scenarios for driving simulation and accident reconstruction. A specific research framework is summarized as shown in Figure 2. Driving simulator and computer simulation are integrated to simulate the whole process of vehicle-pedestrian crash, from precrash to postcrash. On the one hand, driving simulator can simulate the driver’s emergency measures and provide deceleration, the front steering angle, and other necessary parameters of the vehicle for the further simulation by computer software. On the other hand, computer simulation will complete the process until vehicle stops after the collision, to achieve a complete reconstruction of the accident. Through the interaction between the real driver and the virtual scenario of computer, the whole simulation of the vehicle-pedestrian crash could be realized in the laboratory.

Figure 2: Flow diagram of research frame.

According to the research frame mentioned above, the research method includes the following stages.

Stage 1. Find out the road conditions at which traffic accidents most likely happened via the preexisting statistics of traffic accidents, and simulate the scenario based on the virtual road conditions in our driving simulator system. Then recruit volunteer experimenters, conduct the driving simulation experiments, and collect the data of collision kinematic parameters.

Stage 2. Take kinematic parameters collected in stage 1 as input, and use PC-CRASH software to simulate the traffic accidents. Then, establish the function of head’s centroid acceleration with time. Calculate the HIC value based on the data of head’s centroid acceleration as injury indicator.

Stage 3. Compare the HIC value among experimenters who take different emergency measures to examine the relationship between driver’s emergency measures and pedestrian’s injury severity.

3. Real-World Accident Statistics

For designing the traffic scenario in accordance with realistic situation, real accidents data should be surveyed and analyzed. The data source is from the real-world accident cases which occurred in Beijing in recent years and analyzed by the Traffic Safety and Accident Reconstruction Lab of Tsinghua University for the Public Security Traffic Administration Bureau of Beijing. The accident data is obtained from the reconnaissance of public security bureau, the survey by inquirer, the diagnosis by hospital or legal medical expert, and the testing by vehicle inspection department.

Using the methods of statistics, 158 cases of typical car-pedestrian accidents are selected and analyzed by in-depth level, from which the relevant parameters are used to obtain the specific collision form between vehicle and pedestrian and the scene features are listed below. The specific time and weather conditions of the accidents are shown in Tables 1 and 2, in which daytime accounts for about 30% and severe weather, for example, rain, snow, and mist, takes about 8%. The distribution of road contact position is shown in Table 3, where the proportion of road section accounts for about half of the whole number. Table 4 represents the motion characteristics of pedestrians, in which the situation of pedestrian walking across the road accounts for 80%.

Table 1: Accident time.
Table 2: Weather conditions.
Table 3: Road contact position.
Table 4: Motion characteristics of pedestrians and vehicles.

The drivers’ response measures mainly include steering and braking before or after the collision. According to the scene record information of the 88 accident cases, the following classification of driver measures is listed in Table 5, where the braking measure during the precrash takes up about 24%, and the one after the collision takes up about 16%.

Table 5: Braking measures of drivers.

In addition, the statistics results of the vehicle collision speed are shown in Figure 3, which has more evenly distribution from 30 to 70 km/h. About 27% of the accidents have collision speed range of 40–50 km/h, and 20% have collision speed of 50–60 km/h.

Figure 3: Distribution of the collision speed.

According to the statistics and analysis, the specific accident scenario is as follows: while a pedestrian is walking across the road section, a car travels at a fast speed and normal straight direction. It is difficult for the drivers to take effective measures to avoid the pedestrian. Therefore, the experiment scenario should be designed based on the situations of pedestrian walking through the road and car moving straight, and there is a contact chance between car and pedestrian. Some other factors need to be considered in the scenario design, as shown in Table 6.

Table 6: Accident factors and the details.

The scenarios that we designed in this study are based primarily on the statistics of 158 pedestrian accidents cases mentioned above.

4. Driving Simulation Experiment

4.1. Experiment Design
4.1.1. Block Design

As mentioned above, there are two measures drivers may take to avoid collision: the first one is to steer the car (vehicle’s lateral movement) and the second one is to control the longitudinal movement like braking. Therefore, we divide each volunteer experimenter into 3 stages, on stage A that only may steer the car, on stage B that only may brake, and on stage C that should steer and brake at the same time.

4.1.2. Scenario Design

The scenario design fall into two major types, the static scenario design and the dynamic scenario design. The static scene includes road infrastructure, traffic signs, signal lights, trees, buildings, and other circumstances. The dynamic scene includes other vehicles, pedestrians, nonmotor vehicles, and other kinetic objects.

The purpose of tests is to obtain the response characteristics of drivers in emergency conditions and to simulate and reconstruct possible accidents. According to the actual existing conditions from investigation and analysis of the accidents, the scene and the variables are identified and designed as shown in Table 7, including the setup and adjustment for different vehicle speed limits for the drivers, road junctions of the conflict triggered, traffic, weather, visibility, and other factors.

Table 7: Design list of the test variety and conditions.

Among them, test variable 4 can be designed according to the general traffic flow characteristics of intersection or road section, which should be continuous and in accordance with the actual situation. Meanwhile, the adoption section distance of driver can be considered, in order to simulate the accident conflict of driver without any awareness about this. Thus the conflict has the chance to eliminate the driver's preparedness psychological effect.

In each conflict section, there may be other vehicles, pedestrians, nonmotor vehicles, and other traffic participants, who might suddenly enter the driver's vision, to increase the contingency and authenticity of accidents.

Therefore, we designed static scenario as straight road in urban district where the accidents are most likely to happen. The three typical scenarios are shown in Figures 4, 5, and 6. As it is mentioned above, there are 3-stage experiments for each volunteer driver. In order to eliminate effects of scenario differences on reliability of experiments, we establish 3 different accident trigger positions in each scenario (corresponding to stages A, B, and C, resp.), as shown in Figures 4, 5, and 6. Considering more actual situation, the motion directions of pedestrian in these scenarios are different.

Figure 4: Scenario of 1st pedestrian conflict on road section.
Figure 5: Scenario of 2nd pedestrian conflict on road section.
Figure 6: Scenario of 3rd pedestrian conflict on road section.

We also establish a preliminary test scenario, as shown in Figure 7, so as to eliminate the effect of experimenters’ familiarity with the scenario before the start of each formal test. In the preliminary test scenario, there are 4 curve road-sections to adapt drivers to the driving simulator adequately, and the drivers will be asked to apply the brake several times to know the related performance well.

Figure 7: Initial testing scenario for experimenters.
4.1.3. Pedestrian’s Walking Speed

The related study shows that the fast walking of pedestrian is 1.3~1.9 m/s [19], and pedestrians who cross the road usually walk much faster. We finally decide to choose 5 km/h as the speed for pedestrians with an additional consideration of equipment limits combined.

4.1.4. Initial Distance between Car and Pedestrian

In our study, the experiments designed a process that driver could actually collide with pedestrian. Therefore, the initial distance between car and pedestrian could be worked out as shown in Figure 8. We assume the speed of the pedestrian is and then measure out the distance from the side of the road to the middle of the road as . Therefore the time pedestrian takes from the start point to the collision point is , which is the value of divided by . Then we assume the speed of the car is . The initial distance between car and pedestrian, which is assumed as , could be worked out as the value of multiplied by . The results show that these assumptions can match up to our experiments.

Figure 8: Triggle distance design.
4.1.5. Recruit Volunteer Experimenters

Considering the fact that drivers, whose driving years are below 3, have a high risk proportion of occurring accidents and that most of these drivers are no more than 30 years old, our study decide to choose volunteer experimenters between 20 and 30 years old. At last, 7 experimenters are recruited for our study. In addition, in order to eliminate the halo effect of experimenters, we explain the purpose of our experiment as collecting data of the collision times and the deviation of driving speed for the study of steering performance in driving simulator system.

4.2. Experimentation Introduction

In order to make the experiments successful, some special rules for the experiments are considered and established. Before the test, drivers should provide their individual information, such as age, driving years, driving experience, and physical condition. In addition, we suggest the drivers do the following during the test.(i)Please drive and control according to the given speed and do not exceed the limit.(ii)Do not cross the road edge and isolation facilities.(iii)Please try to avoid collisions with other vehicles and pedestrians. In the first conflict scenario, driver should apply steering only. In the second, driver should apply brake only. In the third, driver should apply both brake and steering measures.(iv)Please obey the traffic rules and do not run through a red light.

4.3. Experimentation Process

(i)Start the driving simulation system and debug the experimental equipment.(ii)Inform the experimenters of test specification, let experimenters read experiment introduction, and fill out questionnaires.(iii)Guide experimenters to get into the driving simulation platform and inform them of the precautions in experiment and the way to operate simulation system.(iv)Start with a 5-minute long preliminary test before the normal test.(v)Remind the experimenters of precautions for emergency measures before the start of each formal test and then load the simulation scene and collect the data.(vi)Guide experimenters to get out of the driving simulation system after all three scenes (A, B, and C) were test.

4.4. Experimentation Results

The expected test output includes both qualitative and quantitative aspects. Qualitative contents include the contact characteristics between vehicle and pedestrian or the situation of driver to avoid collision. Quantitative indicators consist of the impact velocity, collision angular velocity, deceleration, steering angle, and relative position of vehicles at the collision moment.

The data are output in log format, and each kinematic parameter is collected with consistent time intervals. In this study, we obtain the data including collision speed, collision deceleration, left front wheel’s steering angle, right front wheel’s steering angle, and driver’s emergency measures for the later computer simulation stages. 7 groups of data were recorded; data with no collision were eliminated. The main results are shown in Table 8, which comprise the collision speed, collision deceleration, and steering angle of front wheels.

Table 8: Experimental results summary.

We can see the various results with different scenarios (A/B/C) and measures taken by various drivers, especially for the steering angle which is within the scope of 0.04~4.61. But there is no distinct rule about the data at present. However, the data will make the obvious influence on the injury of pedestrian in the subsequent computer simulation process.

5. Computer Simulation

After the driving simulation, we obtain the data which represent the driver’s emergency measures. In order to find out how the crash influences the pedestrian’s injury, further we use the data as input to carry out the computer simulation by PC-CRASH software.

The software PC-CRASH we used in our study contains various universal models in its PRO database. In this study, we choose the model of collision between pedestrian and passenger-car and set up the friction coefficient of road surface large enough to match up with the allowed deceleration maximum in our experiments.

We use BMW 3 series in the experiment for this study. Therefore, we first select the vehicle parameters of BMW 3 series in DAT database. Then, three kinetic parameters including deceleration, speed, and front-wheel’s steering angle are also set according to the different scenarios. The parameters of pedestrian’s model (170 cm height, 61 Kg weight, and 5 Km/h speed) are also established to match the built-up database.

The simulating process starts from the collision to separation between pedestrian and car finally to the moment that pedestrian falls to the ground. The data collected from simulator experiments is used as input, and the trajectory of pedestrian and vehicle is simulated as well as the 3D process and acceleration of head centroid. One group of the 7 output results is shown in Figures 9, 10, and 11, respectively. For each figure, A, B, and C represent the first (steering only), second (braking only), and third scenario (steering with braking), respectively.

Figure 9: Experimenter 1-trajectory of pedestrian and vehicle.
Figure 10: Pedestrian’s head centroid acceleration.
Figure 11: Experimenter 1–3D simulation of the crash.

6. Pedestrian Injury Analysis

6.1. Computation on HIC

HIC value is widely accepted indicator for the severity of injuries on head of human body, which is described by where “” is the ratio of acceleration of head centroid to gravity acceleration; ~ represents integral interval for time. The HIC value of the collision simulation results is calculated and shown in Table 9, which represents that the different measures have brought different degrees of pedestrian injuries.

Table 9: HIC calculation results.
6.2. Discussions

The computation results show the “Steering + Braking” to be the most effective measure among three measures in our experiments. It is not only because of its low collision rate but also its low injury severity. The results also show that almost all HIC values in “braking” stage are lower than those of “steering” stage. Furthermore, there are other two statistical results worth mentioning. The first result is that, in the “steering” stage, the bigger the front-wheel’s steering angle is, the higher the HIC values are; however, in the “braking” stage, results show that the higher the deceleration is, the lower the HIC value will be. We would probably discuss the results from the view of energy. First of all, we admit the statement that the more kinetic energy a car has when it collides pedestrian, the more severe the pedestrian’s injury may be. Then we will give a comparative analysis of car’s kinetic energy between “steering” stage and “braking” stage. The car’s kinetic energy in “steering” stage enlarged in that drivers who tried to avoid the collision only by steering would result in an extra rotational kinetic energy adding up to the total kinetic energy; however, car’s kinetic energy in “braking” stage reduced in that drivers who tried to avoid collision by braking would slow down the speed which consequently decreased the total kinetic energy. As mentioned above, the car’s kinetic energy in “steering” stage enlarged; therefore, pedestrians in “steering” stage are injured more severely than inother stages.

7. Conclusion

In this study, based on the in-depth data of 158 pedestrian-vehicle accidents in Beijing, we obtain the realistic accident modality, impact characteristics and road scene, and establish typical virtual scenarios of accidents. Then we carry out driving simulator experiments and analyze the data by computer simulation to find out the relationship between the collision avoidance measures taken by drivers and the injury severity of pedestrians. Through the experiment and simulation, we draw the following conclusions.(1)The way of “braking” is more effective in reducing the severity of pedestrian’s injury than the way “steering” under the same constraints.(2)Under the same conditions, the “Steering + Braking” way is the most effective way among three different emergency measures, in that it has the lowest collision rate and the relatively lower HIC value which indicate its low injury severity when collision.(3)The introducing of driving simulator method into computer simulation could help to take an extra consideration of drivers’ behaviors to simulate a more reliable crash process.

As a result, based on the interaction between real human and virtual scenario, we achieve the research purpose to combine the real driver’s operation and the virtual pedestrian’s motion. However, there are a number of limitations in the research. Firstly, we could augment the samples by recruiting experimenters whose driving years are above 10 years. Secondly, even though the HIC value is a well-accepted indicator for injury severity of pedestrians, we could add up more indicators to reach unbiased conclusions. Thirdly, we could conduct experiments on the collision between car and other vulnerable road users like cyclist. Last but not least, we should consider using other computer simulation software in our future studies.

Conflict of Interests

The authors declare that there is no conflict of interests regarding the publication of this paper.


This work is financially supported by the Opening Project of Key Laboratory of Evidence Science (China University of Political Science and Law), Ministry of Education (2011KFKT01), China. The accident data in this paper are selected from the accident cases that we analyzed for Beijing Traffic Management Bureau.


  1. C. V. Zegeer and M. Bushell, “Pedestrian crash trends and potential countermeasures from around the world,” Accident Analysis and Prevention, vol. 44, no. 1, pp. 3–11, 2012. View at Publisher · View at Google Scholar · View at Scopus
  2. Traffic Administration Bureau of Chinese Public Security Ministry, Statistics Annals of Road Traffic Accident of PRC, 2012.
  3. K. Mizuno and J. Kajzer, “Head injuries in vehicle-pedestrian impact,” SAE Paper 2000-01-157, 2000.
  4. J. A. Stammen, S. Ko, D. A. Guenther, and G. Heydinger, “A demographic analysis and reconstruction of selected cases from the pedestrian crash data study,” SAE Technical Paper 2002-01- 0560, 2002.
  5. L. N. Boyle and J. D. Lee, “Using driving simulators to assess driving safety,” Accident Analysis and Prevention, vol. 42, no. 3, pp. 785–787, 2010. View at Publisher · View at Google Scholar · View at Scopus
  6. K. Hiroaki, K. Akira, H. Atsushi, et al., “Developing a driving simulator and an assessment methodology to estimate a driver’s mental state when distracted,” SAE Paper 2004-01-0233, 2004.
  7. L. N. Boyle, J. Tippin, A. Paul, and M. Rizzo, “Driver performance in the moments surrounding a microsleep,” Transportation Research F, vol. 11, no. 2, pp. 126–136, 2008. View at Publisher · View at Google Scholar · View at Scopus
  8. S. John, M. Ken, L. Eero, et al., “Driver workload effects of cell phone, music player, and text messaging tasks with the ford SYNC voice interface versus handheld visual-manual interfaces,” SAE Paper 2009-01-0786.
  9. M. E. Rakauskas, N. J. Ward, E. R. Boer, E. M. Bernat, M. Cadwallader, and C. J. Patrick, “Combined effects of alcohol and distraction on driving performance,” Accident Analysis and Prevention, vol. 40, no. 5, pp. 1742–1749, 2008. View at Publisher · View at Google Scholar · View at Scopus
  10. B. Ole, M. Knight, B. Walter, and A. Sannier, “Collaborative driving simulation,” SAE Paper 2002-01-1222.
  11. D. Norfleet, J. Wagner, K. Alexander, et al., “Automotive driving simulators: research, education, and entertainment,” SAE Paper 2009-01-0533, 2009.
  12. V. Bhise, E. Smid, and J. Dowd, “ACE driving sumulator and its applications to evaluate driver interfaces,” SAE Paper 2003-01-0124, 2003.
  13. X. Chen, L. Gao, and Z. Wei, “Driver’s braking behavior under emergency,” Journal of Beijing Polytechnic University, vol. 33, no. 6, pp. 603–608, 2007.
  14. J.-H. Kim, S. Hayakawa, T. Suzuki et al., “Modeling of driver's collision avoidance maneuver based on controller switching model,” IEEE Transactions on Systems, Man, and Cybernetics B, vol. 35, no. 6, pp. 1131–1143, 2005. View at Publisher · View at Google Scholar · View at Scopus
  15. H. Steffan, A. Moser, and B. C. Geigl, “A new approach to occupant simulation through the coupling of PC-crash and MADYMO,” SAE Paper 1999-01-0444, 1999.
  16. A. Moser, H. Steffan, and G. Kasanický, “The pedestrian model in PC-crash—the introduction of a multi body system and its validation,” SAE Paper 1999-01-0445, 1999.
  17. R. Fay, R. Robinette, J. Scott, and P. Fay, “PC-Crash and HVE, an overview of similarities and differences,” SAE Paper 2001-01-0505, 2001.
  18. A. Moser, H. Hoschopf, H. Steffan, and G. Kasanicky, “Validation of the PC-crash pedestrian model,” SAE Technical Paper 2000-01-0847, 2000.
  19. Y. Lin, Practical Identification of Car Accident, Communications Press, Beijing, China, 2001.