| Our country's car parc and number of auto drivers continue to increase, and the road traffic safety situation is very grim. Countermeasures like enhance of legislation and enforcement of traffic laws, improvement of road and vehicle safety, increase of people's awareness of traffic safety, promotion of drivers driving skills and decision-making reliability, and so on, have certain significance for decreasing of road traffic safety accidents. But a large number of accident data show that driver is the main cause of traffic accidents. The reliability of driver's decision-making determines the road traffic safety to a great extent. Driver's cognition is the premise and basis of driving decision-making and it has more significant meaning to reduce traffic accident rate than external interference measures such as repairing roads and improving vehicle performance. In order to improve driver's behavior performance and human reliability, reduce and avoid road traffic accidents, it necessary to research the influence of driver's cognitive characteristics on driver behavior decision-making reliability, then analyze the basis of driving behavior decision-making and mechanism of driving error, and it also is a key point and development direction of driving behavior research.In driving field, it is used to describe and explain driving behavior based on probability statistical rules of group behavior and measure driving behavior reliability based on the calculation of human error probability value at present. Existing driving behavior model exist some limitations in describing and explaining driving behavior as follows:descriptive models mainly use static and classified expression of driving behavior and less related to decision-making mechanism; functional models focus on prediction, and most of them are physical or mathematical models which use definite physical meaning of parameters, such as the headway distance, relative velocity or visual angle. Such parameters have high precision but don't accurately reflect actual vehicle road characteristics and driving behavior. While driving human error or reliability analysis either selected variables, such as safety consciousness, accident proneness and attention quality, are often unable to explain unsafe behavior of drivers materially, or related variables are too numerous, so it is difficult to prominent the key factor influence on safety behavior.Driving can be regarded as a process driver operate vehicle to bypass obstacles based on the cognition of relevant information and reach his destination safely. Keeping a reasonable distance with obstacles is the basic goal of driving decision, and it is the key to ensure driving safety. Existing research on distance-control are mainly based on rational behavior and physics laws, and they mainly use theoretical safety distance (TSD) in physics but few pay attention to driver's subjective expected safety distance (ESD). As we know, driver's decision-making behaviors often vary widely even when they face same traffic environment. Vehicle anti-collision alarm device obeys TSD but driver obeys ESD. It could assume that:driver controls actual vehicle distance base on ESD; ESD is the "reference point" of driving behavior decision space; reliability of ESD is the key item of driving reliability.Based on above analysis, on the basis of construction of driving behavior concept model, this paper firstly studies shape and semantic cognitive characteristics of driving obstacles with ergonomics experiment method. Then, it analyzes and verifies existence and individual difference of ESD based on simulation experiment. Finally, it compares the cognitive mechanism of two kinds of driving errors (false and missing) with experimental method of EEG. The main research work of this paper is as follows:A driving behavior decision concept model was built base on ESD, and be described by a 6-tuple. Driving decision is dynamic iterative processes, which take vehicle as starting point and obstacles as endpoint, and continuously carry a dynamic comparison of ESD and PD, then matching the driving behavior decision goal according to the comparison results. Driver behavior decision model can be expressed as DBDM=(A,B,PD,ESD,C,D). In above mentioned 6-tuple, A represents the position of vehicle drove by driver (vehicle starting position), B represents obstacles which need to avoid in driving process, PD represents driver's perception distance, ESD represents driver's expected safety distance, C represents matching process between PD and ESD, D represents driver's decision goals (results).A human reliability concept model was proposed base on ESD. The driving human reliability is regarded as the product of driving dynamics process's stages and link reliability, and reliability of ESD is the key of it. Reliability of ESD (RESD) could be expressed as ESD/TSD ratio, and its values of different ranges could be corresponding to different types of driving human errors (false-alarm and miss).Some regularities of obstacles recognition (shape recognition) under vibration condition were revealed. Experimental test was program by Microsoft Visual Basic 6 to simulate object (obstacle) independent vibration. Experimental result shows:the main effects and interactive effect of vibration frequency and amplitude on obstacle identification efficiency are notable, and identification efficiency decreases exponentially with the increasing of vibration frequency or amplitude. Obstacle characters have significant influence on recognition efficiency, and have notable interaction with direction of vibration; vibration acceleration (x) has a significant main effect on recognition reaction time (y), their relation could be described by exponential function y= 487.382 ×e0.112x vibration acceleration respectively with vibration direction and Obstacle characters have notable interactions.Special obstacle (traffic sign) comprehension effectiveness (semantic cognition) was analyzed. Experimental test was program by E-Prime. Basic types of four kinds of common road traffic signs, similar graphics of them and new graphics were used as recognition objects, and matched with following semantics "prohibition, warning, instruction and prompt". The result shows:new graphics'recognition efficiency and reliability decrease; when matching relation between original graphics and function (semantic) changes, graphics recognition efficiency and reliability reduce because of learning negative migration. Thus, keeping the consistency of traffic sign appearance (form) and function (semantic) could improve its recognition efficiency and reliability to driver. In order to improve traffic signs comprehension effectiveness, it suggests traffic signs with similar functions use uniform shape.Driver's cognitive characteristics about driving safety distance based on reference point were discussed, and the existence of ESD was verified. Car-following behavior was simulated by C# computer programming, and experimental value of car-following distance was used as the representative value of ESD. The experimental result shows that:ESD varies obviously among individual drivers; gender has some influence on ESD; driving type has very significant influence on ESD (P<0.001). ESD is the reference point for drivers to judge whether the vehicle distance is safe, and it is the main cause for differences of actual driving behavior such as car-following distance.Driving error was divided into two categories of "false alarm" and "miss" base on signal detection theory (SDT), and their electroencephalogram (EEG) difference in different cognitive processing stages was compared using delayed matching-to-sample task paradigm and event-related brain potentials(ERP) technology. E-Prime program was used to arrange and present the experimental stimuli and then collect the data of reaction time, and a German BP Company-made 64-channels electroencephalogram recording system was used to collect the EEG signal. The experimental result shows that:reaction time of "miss" is greater than reaction time of "false alarm" (P< 0.05); in information encoding stage, average peak of P300 in Pz electrode point has significant difference between two types of human errors, P300 amplitude of "miss" is significantly greater than P300 amplitude of "false alarm"(p<0.01). The reason of different human error is related to the way and degree of information coding. Amplitude of P300 could be an EEG index to reflect the different processing mechanism of "false alarm" and "miss".In this research, road traffic safety problem is studied from the perspective of reliability of driver's cognition and driving behavior decision-making. The research results could supplement and perfect existing driving behavior model, also help to promote further development of traffic safety and intelligent vehicle field. |
PDF Full Text Request