Driver’s Behavior Pattern And Performance Assessment Method Based On Hierarchical Driving Simulation Experiments

Driving safety is one of the important research points in the field of road traffic. The casualties and property losses caused by road traffic accidents every year are extremely serious, and most of the road traffic accidents are related with the drivers. Therefore, investigation about driver behaviors and performance is an important part of driving safety research. This study aimed to put forward systematic analyses of the driver behavior and performance from the point of driving safety by conducting series of driving simulation experiments. This paper described the process of behavior analysis and driver performance evaluation based on driving simulation experiments from three aspects, which included driving behavior formation mechanism, experimental methods and evaluation methods. Besides, examples analysis and three cases study were conducted. The main contents are shown as follows:(1) Research on the basic theory of driving behavior. This paper focused on the investigation of the cognitive characteristics of driving behavior and formation mechanism of driving risk, and put forward a hierarchical driving behavior model based on the relationship between subjective driving risk and objective driving risk. First of all, driving cognition is the basis of all driving behavior, thus based on SRK (Skill-Rule-Knowledge) cognitive model, the descriptions of driving behavior and its disadvantages were introduced, so as to highlight the important role of the subjective risk judgment in the driving behavior. The driving task demand and the driving ability are important factors of driving behavior. Based on the TCI model, this paper introduced the relationship between these two factors and difficulty of driving tasks (objective driving risk), and described the deficiency of TCI (Task Capability Interface) model. Based on the relationship between subjective driving risk and objective driving risk, the driving behaviors were divided into three risk levels according to objective driving risk, and a hierarchical driving behavior model was proposed in this paper. Low-risk driving behavior referred to the basic vehicle control. Mid-risk driving behavior referred to the dynamic decision-making. High-risk driving behavior referred to the emergency response. This model can describe the inherentcharacteristic of driving behavior based on accident risk, and guide driving simulator experiment design to detect driving behavior and evaluate driving performance.(2) Research on the method of driving simulation scenarios design. Due to the high-risk driving scenarios, the driving simulator is considered to be an ideal tool for the analyses and evaluation of the driver’s behavior performance. This paper put forward the design method of simulated scenarios based on the hierarchical model from three aspects including the risk classification of driving simulation scenarios, modular construction of road scenarios, and standardized design of driving experiment scenarios. First of all, the scenarios were classified based on the hierarchical driving behavior model. According to the risk level, the scenarios were extracted from the previous studies, and the design characteristic parameters of each scenario were confirmed. Then, the module construction method and the standard design method of experimental scenarios were put forward based on the driving simulator. The road scenarios could be generated quickly using the module library. The scenario characteristic parameters were set based on the dynamic model library, and last the experimental scenarios design could be finished by adding the intervening factors.(3) Research on the evaluation method of drivers’performance. According to the experimental results, the driver performance evaluation of a single driving scenario were assessed by extracting the evaluation index, based on which, the drivers’ performance could be also assessed specific to driving scenarios tree. In this paper, the characteristics of experimental data and the post processing methods were introduced first. The selection rules and extraction methods of evaluation index were provided on the basis of index classification. Then, the evaluation method of evaluation index for a single driving scenario was provided based on experiment statistical theory. Finally, the evaluation model of driving scenarios tree was provided based on the driver’s performance evaluation results for a single scenario, and case studies for evaluation method of a single scenario and the evaluation model of driving scenarios tree were carried out respectively.(4) Research on the effect evaluation of red light violation warning on driving behavior and performance. In this part, a mid-risk scenario of yellow light go/top at signal intersection was designed with red light violation warning as the intervention factor. Drivers could receive audio warning information through the vehicular audio system when the traffic signal changed. By comparing the driving behavior and assessing drivers’performance under warning and without warning conditions, the effectiveness of red light violation warning technology was analyzed. And the application of mid-risk single scenario assessment method was provided from three aspects which were experiment design, driving behavior analysis and drivers" performance evaluation. Results showed that red light violation warning technology can help drivers to better conduct go/stop decision-making behavior and reduce the risk.(5) Research on the effect evaluation of signalized intersection crash avoidance warning on driving behavior and performance. In this part, a high-risk scenario of red light running vehicle avoidance at signalized intersection was designed and the crash avoidance warning was the intervention factor. Drivers could receive audio warning information through the vehicular audio system before they found the red light running vehicle. By comparing the driving behavior and assessing drivers’performance under warning and without warning conditions, early warning and late warning conditions and warning conditions with and without directional information, the effectiveness of signalized intersection crash avoidance warning was analyzed and the best warning condition was found. The application of high-risk single scenario assessment method was provided from three aspects including experiment design, driving behavior analysis and drivers’performance evaluation. Experiment results showed that signalized intersection avoidance warning system can help drivers to better take brake actions and reduce the risk. In addition, early warning was the best warning condition while the effect of warning with/without directional information on driving behavior was not significant.(6) Research on the effect evaluation of handheld cell phone conversation on driving behavior and performance. In this part, a conversation through handheld cell phone was the intervention factor. Scenario tree was constituted by selecting scenarios of different risk levels. Firstly, a single scenario driving performance evaluation score can be obtained by comparing driving behavior and performance with/without cell phone conversations. Secondly, based on the single scenario driving performance evaluation score, a comprehensive driving performance evaluation was scored for scenario tree. Finally, the influence of handheld cell phone conversation on driving safety was analyzed and the application of multi-scenario (scenario tree) assessment method was provided from the following aspects:experiment scenario selection, driving behavior analysis and drivers’performance evaluation. Results showed that handheld cell phone conversation had a negative effect on driving behavior and increased the driving risk.