20 April 2018, Volume 31 Issue 4
    

  • Select all
    |
  • Influence of Operating Types of Smartphone on Drivers' Car-following Behavior（in English）
    GAO Yan, LUO Yi, YOU Zhi-dong, ZHOU Mu-xiong, HAO Hui-long
    China Journal of Highway and Transport. 2018, 31(4): 1-9,58.
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    To explore the influence of different operating types of smartphones on drivers' car-following behaviors, the operating types were divided into eight classes according to their function type, and usage pattern and driving simulation experiments were conducted using the driving simulator. The car-following speed, headway, time headway, lateral deviation distance, and steering wheel angle were chosen to describe the vehicle's longitudinal and lateral running status and the entropy of distribution of the fixation point, distribution ratio of the fixation point, fixation duration, saccade frequency, saccade duration, blink frequency, and frequency duration were defined to describe the driver's vehicle operation characteristics. In addition, the driver's visual characteristics were analyzed, and ANOVA was used to verify the validity regarding whether such indicators can measure a driver's car-following behavior. The results show that specific smartphone operations could have a significant influence on the vehicle's running indicators and the driver's visual indicators. A grey relational analysis (GRA) was applied to quantify the influence of these eight operating types of smartphone on the driver's car-following behavior, and a comparison of the operating types with the same function related to the content and type of distraction, and the action time of the operation, was conducted. The results show that, when driving, the influence of these eight typical distracting smartphone operations on the driving performance are in the following order from big to small:replying to text messages, reading text messages, hand-held answering, replying to voice messages, touchtone dialing, reading voice messages, voice dialing, and hand-free answering in which the typing and reading of text messages are of greater influence than other smartphone operations. The results of this study can help drivers make clear the hazards of different operating types of smartphones on driving safety.
  • Eye Gaze Characteristics of Overtaking Process in Very High Altitude Area
    WANG Fan, CHENG Hui-ting, LIU Jian-bei, LI Yan
    China Journal of Highway and Transport. 2018, 31(4): 10-17.
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    To obtain the eye gaze characteristic of an overtaking process in very high altitude areas that are higher than 3 500 m, the eye movement data from real vehicle experiments at the Qinghai-Tibet Highway was recorded using eye-tracking glasses. The fixed durations within the seven areas of the drivers' vision and the fixed transitions during the four phases of the overtaking process were collected. The Markov process was utilized to obtain the steady-state fixation distribution of every overtaking phase for the four ranges with various altitudes. The connectivity indicator was defined as the maximum eigenvalue of the correlation coefficient matrix. The differences in the drivers' fixation between various overtaking phases were analyzed using the correlation coefficients between the corresponding features and connectivity indicators. The results of the studies indicate that the fixation rates on the current lane and target lane at a very high altitude area are 61.8% and 26.4% respectively, which are the drivers' most interested area; the connectivity indicators of the four overtaking phases are 3.315, 2.934, 3.102, and 3.794, which illustrate that the purpose phase has the most significant otherness, and the car following and returning phases have similar fixation mode. The correlation coefficients obtained from the drivers that be in the very high altitude area for the first time are 1.89 times larger than those of the local drivers, indicating that they are more cautious during driving and are more discrete on their fixation. The correlation coefficient of the eye gaze characteristics of the area with an altitude higher than 4 400 m is 39.3% lower than other areas investigated, which leads to a different eye gaze pattern; the drivers at the very high altitude area will primarily focus on the current lane, while the drivers at the plain or hilly areas will have a higher fixation on the target lane, implying that the drivers at the very high altitude area should be more vigilant to the driving environment. The method proposed can determine the eye gaze characteristics of drivers in very-high-altitude areas. The results can be utilized in the fields of road infrastructure design and active safety prevention.
  • Effects of Driving Style on Driver Behavior
    HOU Hai-jing, JIN Li-sheng, GUAN Zhi-wei, DU Hai-xing, LI Jing-jun
    China Journal of Highway and Transport. 2018, 31(4): 18-27.
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    To reveal the influence laws of driving style on driver behavior and to select the driving style characteristic parameters, the quantitative analysis of the driver behavior data of different style drivers was performed, which included the data in situation awareness and operational layers during driving. First, a questionnaire survey of driving style was conducted on 18 non-professional Chinese drivers based on the driver behavior questionnaire; additionally, principal components analysis and K-means clustering were used to quantify the driving style and classify the 18 participants into three groups:cautious, normal, and aggressive drivers. Subsequently, the 18 drivers drove in the highway traffic environment on a driving simulator equipped with a SmartEye system; meanwhile, the visual characteristic parameters and driving performance parameters were acquired synchronously, and a judgment sampling method was used to select the samples according to the driving styles and driving maneuvers (which included lane changing intention and lane keeping). Subsequently, 810 effective samples were obtained. Finally, the analysis of variance method was used to analyze the significant difference between the different driving styles in gaze behavior, saccade behavior, as well as lateral and longitudinal control characteristics during different maneuvers, followed by the extraction of parameters that have significant differences between different styles as the driving style characteristic parameters. The results show that the aggressive drivers exhibit less attention to the surrounding environment, worse stability in lateral control, and more frequent emergency accelerating and braking. Additionally, the differences between different driving styles are significant in those parameters:gaze times to the rearview mirror during the intention time window(p=0.002), steering entropy(p=0.04), accelerator pedal position(p=0.01), and brake pedal position(p=0.02). Accordingly, these four parameters are selected as the characteristic parameters of the driving styles.
  • Driver's Minimum Required Visual Attention in High Speed Car-following （in English）
    LIU Zhuo-fan, FU Rui, MA Yong, YUAN Wei, CHENG Wen-dong
    China Journal of Highway and Transport. 2018, 31(4): 28-35.
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    To avoid the limitations of current research methods for driving distraction, we explored the minimum required visual attention for safe driving under car-following condition on highway from the perspective of attention demand. An experiment was carried out on a driving simulator to record the visual behavior and vision occlusion data of 26 drivers under both normal and occluded driving conditions. An analysis of occlusion behavior and the difference of the visual behavior under the two driving conditions were conducted. The results show that safe driving can be achieved at about 35 m without sampling traffic information under car-following conditions on highway. The occlusion distance was independent of speed and can be used as an indicator of attention demand. Although there were individual differences in occlusion distance and occlusion frequencies, the drivers' overall occlusion percentage remained relatively stable. In combination with the results of visual behavior analysis, the drivers' spare capacity under car-following conditions on highway was mainly used to observe the forward road and other areas. Specifically, the drivers' glance distance to forward road and other areas were significantly shorter under conditions of vision occlusion, whereas the glance frequency did not change significantly. In addition, the drivers could sample sufficient information within only about 25m, which indicates that the glance frequency is more important for information sampling. In consideration of individual differences, the distribution of the minimum required visual attention was primarily explored. In addition, the pattern was set based on the results:the forward road, speedometer, and mirror need to be observed within at least less than 20-60 m (1.0-2.8 s), 80-220 m (4.1-8.6 s), and 140-300 m (6.7-13.5 s), respectively. In addition, the drivers could have a chance to occlude themselves every 50-200 m (2.5-9.1 s); however, the occlusion distance should be less than 43.7 m (2.4 s). The findings from this research can be utilized in context-sensitive distraction mitigation systems and better design of human-machine interface.
  • Influence of Distraction on Driver's Reaction Time to Traffic Conflicts（in English）
    LI Peng-hui, HU Meng-xia, ZHANG Wen-hui, LI Yi-bing
    China Journal of Highway and Transport. 2018, 31(4): 36-42.
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    To investigate the effect of driver distraction on drivers' reaction time to traffic conflicts, we used a driving simulator to construct two typical traffic conflicts in city-road traffic environments:lateral pedestrian crossing conflict and longitudinal rear-end conflict. Three types of distracted tasks, including cognitive, visual and text (cognitive + visual combined) distraction tasks, were designed. In the conditions of different speeds, following time headway and lead vehicle deceleration, brake reaction times (BRT) from 30 participants responding to the conflicts were collected, and modeled by the repeated measures general linear model and the linear mixed model. The results show that cognitive distraction produces a 0.09 s increment in the BRT to lateral pedestrian crossing conflict compared to baseline, whereas no significant effects on the BRT to longitudinal rear-end conflict are observed; visual distraction and text, however, increase the drivers' BRT to both lateral pedestrian crossing (0.31 s increment for visual distraction and 0.27 s for text) and longitudinal rear-end (0.47 s increment for visual distraction and 0.38 s for text) conflicts. In addition, in the longitudinal rear-end conflict, with the increase in situational urgency (the increase in lead vehicle deceleration, the decrease in following time headway, and the increase in own-vehicle speed), drivers' BRT significantly decreased. Our results suggest that, driver distraction increases drivers' BRT to traffic conflicts, thus causing traffic accidents. Specifically, cognitive distraction primarily increases BRT to lateral conflicts, whereas visual-related distractions increase BRT to both lateral and longitudinal conflicts. Visual distraction leads to higher increment in BRT than cognitive distraction, indicating that visual distraction has larger effects on driving safety than cognitive distraction.
  • Driver's Distraction States Identification in Simulating Driving Environment
    ZHANG Hui, QIAN Da-lin, SHAO Chun-fu, CHEN Qing-min, SHAN Qing-chao
    China Journal of Highway and Transport. 2018, 31(4): 43-51.
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    To explore the recogonition method of driver distraction states, an identification model of driver distraction states was established. Therefore, an identification model to explore the recognition method of driver distraction states was established. First, this paper describes the design of a distracted-driver driving experiment conducted to collect experimental data, which includes the driving performance and driver's eye movement features during normal and distracted driving conditions, and the establishment of a subset of alternative indicators for driver distraction states. Second, support vector machine-recursive feature elimination (SVM-RFE) was used to rank these indicators, and the importance sequence of indicators was obtained. Third, some of the importance indicators were successively selected as the input of a back propagation (BP) neural network. Owing to its global search performance, the genetic algorithm was used for each BP neural network classification model for optimizing the weights and thresholds. The optimized BP neural network model was applied as a new weak classifier, and using the AdaBoost algorithm, many of these weak classifiers were composed into a strong classifier model. The Adaboost-GA-BP combined model was used for an identification of the driver distraction states. Finally, the performances of the models under different indicators were calculated using the experimental data to determine the optimal discriminant indicators, and the model was then validated and evaluated. The results show that the optimal features of the model are the 14 indicators of the importance sequence, and that the recognition accuracy of the driver distraction states is 95.09%. Compared to the BP, GA-BP, and Adaboost-BP models, the Adaboost-GA-BP model is superior to the other three methods in terms of accuracy, precision, recall, F1-measure, and ROC curve. This model can effectively determine the driver distraction state, which can provide supporting data for a driver distraction warning system and a control strategy.
  • Effect of Information Stimulus on Driver's Vigilance Improvement
    WANG Lei, YANG Chen-xuan, LU Qiao-zhen, ZHU Tong
    China Journal of Highway and Transport. 2018, 31(4): 52-58.
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    To study the improvement of a driver's vigilance under various information stimuli, a driving simulation test was constructed. Drivers were asked to drive the primary vehicle to follow the front vehicle and report the case when the braking deceleration of the vehicle ahead exceeds the threshold. The two-hour experiment was divided into six parts and some of them contained three sets of different levels of motivation, including no incentive, music incentives, and information incentives. The heart rate and the brain wave α and θ, respectively, of the drivers were recorded by MP 150. Meanwhile, the hit rate and false negative rate of the driver's judgment were also recorded. Before and after the experiment, the SOFI fatigue measurement table was filled out by the participants separately. The results of the test showed that under the level of non-incentive, the driver's vigilance decreased gradually over time. In addition, the hit rate decreased from 84.7% to 65.2%, the heart rate also decreased simultaneously but the brain waves α, θ increased. Additionally, under the level of music incentives, the driver's vigilance and brain waves α, θ maintained a relatively stable level. This incentive slowed down the decline of the vigilance to some extent. Moreover, under the level of information incentives, the signal hit rate increased significantly. Although the change in the brain wave α was not obvious, the brain wave θ increased. In addition, the heart rates in all of the groups above decreased over the experimental time. It is concluded that the appropriate music incentives can maintain the drivers' vigilance at a stable level. Furthermore, the information incentives can raise the driver's awareness to perceived risks during a particular time period. The results of this research are valuable for the design and management of a driving-assistant safety system.
  • Fatigue Characteristics in Drivers of Different Ages Based on Analysis of EEG
    PEI Yu-long, JIN Ying-qun, CHEN He-fei
    China Journal of Highway and Transport. 2018, 31(4): 59-65,77.
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    To determine the status of fatigue and the speed of fatigue accumulation in drivers of different ages during driving, we compared the difference in fatigue generation and change to obtain the optimal driving time for drivers of different ages, and a natural driving experiment was designed in our study. The electroencephalogram (EEG) data of drivers was collected through Physio, a physiological multichannel instrument. The drivers were inquired by the subjective test method simultaneously. MATLAB was applied to denoise the collected EEG data and the average densities of power spectrum of the α wave, β wave, and θ wave in each period were calculated through integration. Subsequently, the EEG indexes, R(α/β), R(θ/β), and R(α+θ)/β, were obtained. The one-way analysis of variance (ANOVA) was performed by comparing them with the driving time using SPSS. R(α+θ)/β was chosen as the characterization index of driving fatigue through sensitivity judgment. The R(α+θ)/β values in drivers of different ages were averaged and were fitted linearly with the driving time such that the impact of the driver's age on the speed of driving fatigue accumulation can be analyzed. The t tests of paired samples were performed on R(α+θ)/β in each period during driving combined with the subjective inquiry results; the optimal driving time in drivers of different ages was obtained. The result shows that the speed of fatigue accumulation of young and middle-aged drivers within 0-1.5 h is relatively slow, while that of the elderly drivers is fast. However, that of the young drivers is the fastest within 1.5-3 h, while that of the middle-aged drivers is the slowest. The optimal driving time of the elderly, middle-aged and young drivers is 60-75 min, 120-135 min, and 105-120 min. Driving experience, physical strength, the energy in drivers of different ages, and the disturbance of external environment are the important factors that affect the speed of fatigue accumulation. The experiment results verify the effectiveness of using R(α+θ)/β as the driving fatigue characterization index, which will help to afford scientific evidence to set a safe driving time for drivers of different ages.
  • Effect of Slope and Velocity on Drivers' Heart Rate Growth in Urban Underpass Tunnel Longitudinal Section
    FENG Zhong-xiang, YANG Miao-miao, MA Chang-xi, ZHANG Wei-hua, DU Ying-jie
    China Journal of Highway and Transport. 2018, 31(4): 66-77.
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    To study the change in physiological and behavioral characteristics in the longitudinal section of an urban underpass tunnel, 22 drivers were selected to performan experiment at off-peak hours from 5:00 to 7:00 in the morning, wherethe traffic conditions are almost the same in the longitudinal section of the urban underpass tunnel. An MP 150 multiconductivity physiological recorder and the speed acquisition equipment were employed to record and analyze the drivers' heart rate and speed. Significant differences in the experimental data were tested using the one-way analysis of variance (ANOVA) method, and the drivers' heart rate growth by the law of the urban underpass tunnel longitudinal section affected by velocity and slope was analyzed. Subsequently, the quantitative model of the drivers' heart rate growth, slope, and velocity was established; further, the relationship between slope, speed, and the driver's heart rate growth rate was quantified. The sensitivity analysis of two independent variables (slope and velocity) in the model was performed by the single factor sensitivity analysis method.The results show that when the driver was driving on the urban underpass tunnel longitudinal section, a statistically significant difference was found among different slope ranges on the velocity and the drivers' heart rate growth;as the slope increased, both the velocity and the drivers' heart rate growth would first increase and then decrease. In both the uphill and downhill sections of an urban underpass tunnel, the velocity is at the maximum in the 3.5%-4.0% slope range. On the uphill section, the drivers' heart rate growth reached the maximum in the 3.5%-4.0% slope range; however, on the downhill section, the drivers' heart rate growth reached the maximum in the 4.0%-4.5% slope range. The drivers' heart rate growth on the tunnel downhill is significantly higher than that on the uphill, indicating that drivers are more nervous while driving downhill. Finally, the driver's heart rate growth is more sensitive to the higher slope higher than to the velocity.The slope change is more likely to cause the driver's heart rate growth rate to change, and the driver's psychological stress level is affected more by the slope.
  • Transient Analysis of the Driving Behavior in the Dumpling Fog Sections of Expressway Based on Catastrophe Theory （in English）
    LI Xiao-lei, TANG Bo-ming, ZENG Chao
    China Journal of Highway and Transport. 2018, 31(4): 78-87.
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    To reveal the mechanism of an accident caused by the rear end of multiple vehicles in the dumpling fog environment, it is necessary to analyze the driving behavior of the driver in the fog section. By clustering analysis, It was found that the dumpling fog section length, the time period of dumpling fog, and the frequency of dumpling fog have catastrophic characteristics. Therefore, the transient analysis of the driving behavior in a dumpling fog environment in the expressway should take full consideration of its mutagenicity. By considering dumpling fog sections with the visibility interval of (0, 1 500 m] and the fog segment length of less than 4 km as the study object, a transient model of the driving behavior based on the visibility, visual area, and driving speed was constructed, and the transient analysis of the driving speed and gazing region of the vehicle after the fog section of the league was performed. The different driving behaviors in the dumpling fog section of an expressway under the same visibility with induced fog lamps and non-induced fog lamps were analyzed simultaneously. The analysis results show that, in the process of entering and exiting the dumpling fog section from a fog-free section, the driving behavior can be divided into eight stages:detection, confirmation, emergency braking, accelerated braking, slow braking, speed retention, micro-acceleration, and rapid acceleration. The acceleration variation in these stages leads to the increase in the discrete value of the speed in the dumpling fog section of the expressway. After adding the fog lamp on both sides of the dumpling fog section, the missing visual information caused by the dumpling fog is compensated, the driver's visual area diameter and sight distance become larger, and the amplitude of deceleration becomes smaller, which helps to reduce the speed disturbance and the abrupt effect, while improving the safety of the vehicles in the dumpling fog section.
  • Empirical Analysis of Risky Driving Behavior During the Phase Transition Intervals at Signalized Intersections Based on Vehicle Trajectory Data
    TANG Ke-shuang, TAN Chao-peng, ZHOU Nan
    China Journal of Highway and Transport. 2018, 31(4): 88-97.
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    Based on the vehicle trajectory data during the phase transition intervals at an intersection, the risky driving behaviors that may occur within these intervals were identified and analyzed individually based on the driving state respectively. Through video capturing and image processing, high-precision vehicle trajectory data, for a total of 312 single vehicle trajectories from three intersections along the Cao'an road and 449 following vehicle trajectories from the Siping-Dalian intersection, were obtained. With the empirical analysis of the risky driving behavior during the phase transition intervals at the intersection, the speed, acceleration, deceleration, jerk peak-to-peak jerk, time to collision (TTC), and other parameters were used to study the characteristic and type of risky driving behaviors within these intervals. For the vehicles traveling in a single state, after a stop-go classification, the risky driving behavior of the stopping vehicle is classified as one of three categories:emergency deceleration, enhanced deceleration, or continuous urgent deceleration, according to the deceleration, jerk, and peak-to-peak jerk. The risky driving behavior of the passing vehicle is classified into one of four categories:red violation, overspeeding, radical acceleration, and continuous high speed, according to the time through the stop line, speed, and acceleration. For the vehicles traveling in the following state, considering the different combinations of leading and following the vehicle's stop-go decision, the risky driving behavior was analyzed with the occurrence of rear-end crash risk according to the TTC in five time intervals (0.12 s). In accordance with the identification of risky driving behavior types, the internal relation of the key parameters and risky driving behaviors were discussed. The identification result shows that 17 percent of the single vehicles have risky driving behaviors, in which 53 percent are emergency deceleration, and 19 percent of the following vehicles are in risk, where the proportion of stopping vehicles is more than two times higher than passing vehicles. The results of the study can be further applied to the parameter calibration reference of a driving behavior model, an intersection safety evaluation method based on the vehicle trajectory, proactive safety control strategy, etc.
  • Effects of Combined Alignments of Mountainous Freeways on Lane Departure
    CHEN Yi-xin, WANG Xue-song
    China Journal of Highway and Transport. 2018, 31(4): 98-104.
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    The objective of this paper is to examine which geometric design characteristics, including combined alignments and adjacent segments, have significant influence on the probability of lane departure, which are not currently available and can help to reduce side-swipe, rear-end or single-vehicle run-off-road crashes through a better road design. The driving scene was reproduced in virtual reality by the exact road design parameters and roadside elements. With a driving simulator study, lane departure of passenger car data and the road alignments at the current segment, upstream and downstream adjacent segments were calculated. Along the driving direction, lane departures were classified into lane departure to the left and lane departure to the right with the reference category of lane keeping. Since lane departure events by driver are correlated, a two-level Logit model was employed to examine the contributory factors from drives and road alignments. The results indicate that lane departures are more likely happened on the curve compared with straight line. Drivers usually tend to keep to the inside of the curve. The difference between maximum and minimum horizontal curvatures and the average vehicle speed within the 300 m adjacent upstream alignment are found to be positively influenced lane departure. Drivers are more cautious on the downward slope (S<-2%) or upward slope (S>2%), which can reduce the probability of lane departure. Factors of driver-level random effects in the model results indicate that 8.8% of the total variation in the lane departure to the left in relation to lane keeping was accounted for by the drivers and 25.6% of the variation in the lane departure to the right was also accounted for by the drivers. These results can assist engineers in designing safer road alignments.
  • Establishment and Application of Sight Distance Computing Model Based on Support Vector Regression （in English）
    CHEN Yu-ren, FU Yun-tian, WANG Fan
    China Journal of Highway and Transport. 2018, 31(4): 105-113.
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    To inspect the sight distance of mountainous expressway tunnels more effectively, a model for sight distance calculation based on the support vector machine was established and applied to the inspection of sight distance in an actual tunnel. First, the visual lane model of center line was established based on vehicle video images, from which six shape parameters were extracted, including the length and curvature of the visual curve. A significant correlation was found among the six parameters through correlation analysis. The principal component analysis method was used to synthesize the six parameters into two irrelevant components. Next, a support vector regression model was established to compute the sight distance using the two components. After training and testing, the mean square error of the model was 0.071 8, and the squared correlation coefficient was 0.359. Subsequently, the support vector regression model was used to inspect the sight distances of 10 tunnels from the Xinchang to Baihe section on the Shangsan expressway in Zhejiang Province, in which the sight distance was found to be insufficient in the Ma'aoling tunnel (south to north) and Longshan tunnel (north to south). Comparing the vehicle video image data of different tunnels, defects in the design of the visual environment were found in the two tunnels, particularly in speed controlling and visual information supplying. The result shows that the horizontal markings with colorful pavements produced a good speed control effect at the entrance of the tunnel, which reduces the demand for sight distance. The flashing contour marking in the tunnel and a green belt near the tunnel exit could enhance the driver's visual perception, thereby reducing the duration of "black hole" and "white hole" and provide a more adequate sight distance. The two methods contribute to the balance of supply and demand of sight distance.
  • Cycle Length Design Model of Edge Rate Speed Control Line on Highway Straight Sections（in English）
    LIU Bing, WU Dong-ping, ZHU Shun-ying, WANG Hong, XIA Jing, JIANG Ruo-xi
    China Journal of Highway and Transport. 2018, 31(4): 114-121,138.
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    To explore the influence of the edge rate speed control line cycle length on the speed control behavior, and to build a cycle length design model, experiments were repeatedly carried out on the straight sections of four highways by controlling the edge rate speed control line cycle length at 16, 8, 4, 2, 1 m. The speed data were collected, the temporal frequency corresponding to every single vehicle was calculated based on the definition of the edge rate, and the relationship between the temporal frequency and acceleration was determined. On this basis, considering the relationship among the accident rate, accident intensity, and vehicle speed index, and combining the characteristics of the road speed distribution, the cycle length design model and the corresponding algorithm of the edge rate speed control line were established. It was found that the temporal frequency is an internal factor of the cycle length affecting the speed reduction. When the temporal frequency was between 10-18 Hz, the drivers consistently braked, and the reduction in speed increased over the temporal frequency, whereas when the temporal frequency was equal to 18 Hz, the most consistent deceleration effect occurred, and significant and stable linear relationships between the temporal frequency and acceleration were shown during the four test sites. When the temporal frequency was between 1-10 Hz, the acceleration corresponding to every temporal frequency was less than zero. When the temporal frequency was greater than 18 Hz, the mean acceleration corresponding to every temporal frequency was less than zero; however, the reduction in speed was significantly less than its counterpart when the temporal frequency was 18 Hz, and the control speed effect was inferior. Accordingly, with the aim of minimizing the standard deviation in speed and keeping the top speed at less than the safe speed, and considering the upper limit of the temporal frequency, the relationship between the temporal frequency and acceleration, the relationship between the end point and the starting speed, and the maximum length of the edge rate speed control line as constraints, a cycle length design model of the edge rate speed control line was established. This model can be used to optimize the cycle length of the edge rate speed control line and improve the speed control effect.
  • Research on Light Environment Improvement Framework of Highway Tunnel Based on Visual Guidance（in English）
    DU Zhi-gang, XU Wan-wan, XIANG Yi-ming
    China Journal of Highway and Transport. 2018, 31(4): 122-129.
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    To solve the contradiction between energy saving and traffic safety with regard to the lighting environment of a highway tunnel, the problems and existing improvement methods of such an environment are analyzed, based upon which, a simple sorting method is adopted to compare the sight guidance, protection, anti-skid, shading, and lighting facilities in terms of economy, durability, and construction difficulty. Moreover, the safety demands of highway tunnel drivers are analyzed according to accident causes. The results show that a visual guidance facility that can reconstruct the visual reference system of a highway tunnel is the first choice for optimization of the lighting environment. In a tunnel section, such a guidance system can effectively control the driver's visual capabilities, meet the driver's safety demands, and reduce poor driving behaviors and traffic accidents. Safety demands a sense of speed, distance, direction, and location. Retro-reflective visual guidance facilities are used to construct a safety visual reference system consisting of multi-frequency, multi-dimension, multi-shape, multi-color information, which can strengthen the local brightness and contrast in a tunnel and improve the driver's sense of speed, distance, direction, and position. A sight guidance system based on safety is suitable for a highway tunnel with low illumination. For low-grade highway tunnels without lighting, a guidance system can replace the lighting facilities. For high-grade highway tunnels with lighting, a guidance system can assist with the lighting facilities. Such a system can help realize a unification of energy saving and traffic safety in a tunnel environment.
  • Analysis of Variable Information Sign Location Based on Potential Bottleneck Recognition（in English）
    WANG Jian-jun, JIN Shu-xin, REN Yu-yue
    China Journal of Highway and Transport. 2018, 31(4): 130-138.
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    To reduce the impact of traffic accidents and traffic congestion on city traffic, the planning point of variable message signs (VMSs) must be scientifically and reasonably determined, and its traffic guidance function should be effectively utilized. With traffic diversion as the precondition of a VMS location, we use the fragile degree of a road section to identify potential bottlenecks. First, definite sections of the VMS location are combined with the affected area of the VMS. Next, the statistical data of the drivers' memory of traffic signs are cited to obtain the attenuation law of traffic signs information with distance, which contribute towards information utilizations. Using the network analysis module of the ARCGIS software to obtain a bypass path, which contributes towards detour utilizations, the redistribution path between nodes starting from the VMS candidate section is solved by an improved user equilibrium model to determine the network's total travel time. Finally, the optimization objective is to maximize the information utilizations, maximize the detour utilizations, and minimize the total travel time of the network. Further, the optimization model of the VMS layout is established using the cost of the VMS as a constraint, which is solved by the fast classification of non-dominant genetic algorithm (NSGA-Ⅱ). By taking the Xi'an city road network as an example, the results show that the optimization schemes of the VMS layout points are obtained to verify the feasibility of the optimization model and algorithm. The more the number of VMS sets, the greater is the overall information utilizations and detour utilizations of the Pareto solutions, and the more is the linear decrease in the overall trip time. The reasonable quantity of the VMS can result in stable Pareto solutions.
  • Optimal Setting Methods of Complex Diagrammatic Guide Signs on Urban Expressway
    HUANG Li-hua, ZHAO Xiao-hua, LI Yang, RONG Jian
    China Journal of Highway and Transport. 2018, 31(4): 139-146.
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    To find the optimization setting methods of complex diagrammatic guide signs, three types of typical complex diagrammatic guide signs of an urban expressway were chosen, and five setting methods (the status and four types of optimization method) were developed in this study to conduct research on an optimization method for complex diagrammatic guide signs. Based on a driving simulation experiment, nine evaluation indexes were selected from four levels, the running state, handling behavior, driving efficiency, and subjective feeling. The grey premium evaluation method was applied to evaluate the effectiveness of different setting methods of complex diagrammatic guide signs. The results show that the status effectiveness of three types of diagrammatic guide signs has not reached the optimal level, and most of them in the middle and lower levels need to be improved. The optimal setting methods of diagrammatic guide signs 1, 2, and 3 are simplified, repeated, and repeated, respectively. The optimal setting methods of complex diagrammatic guide signs can effectively improve the driving behaviors. In the four optimization methods, the influence of the advanced setting method is more stable, which influences the diagrammatic guide sign to reach the general level or higher. When the exit department coordinates with the ground text setting, the guiding effect of three diagrammatic guide signs lies at the general level or lower, which is barely satisfactory. This research reveals that complex diagrammatic guide signs with more than two exits have poor effects on driving behavior. Related management departments should strengthen the optimization setting for this type of sign. Different complex diagrammatic guide signs have different effects on driver cognition and behaviors. To improve the driving behaviors effectively and guide the drivers to find the destination exits quickly, different optimization setting methods should be carried out for them. In practical traffic engineering applications, the setting up of complex diagrammatic guide signs can be considered in advance when the economic cost, spatial conditions, and other factors are limited. It is not recommended to set the ground text at the exit department. This method does not improve the driving behavior under the influence of diagrammatic guide signs.
  • Visibility of LED Active Luminous Traffic Signs（in English）
    LU Jian, XU Tao, PENG Yi-chuan, LIU Gan
    China Journal of Highway and Transport. 2018, 31(4): 147-155.
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    Driving tests were performed to study the visibility of signs using LED dot-matrix displays, LED transflective displays, and conventional retro reflective screens, to provide a reference for the design and performance of road traffic signs. The effects of environmental luminance, flashing frequency, and lighting forms were analyzed. Variance analysis was used to study the significant effects and interactions of all factors on the visible distance. The paired sample t-test was performed based on the data collected from driving simulators to compare the visible distance between signs using transflective displays and retro reflective screens. In addition, the optimal luminance interval for signs using transflective displays under different environmental luminance was analyzed. The results indicate that the visible distance for LED active luminous signs is greater than that for signs using retro reflective screens by 2-4 m in the afternoon (200 lx). The visible distance for signs using LED dot-matrix displays and LED transflective displays increases by 39.7% and 80.4%, respectively, at night. The visibility of LED active luminous signs does not improve with flashing light irrespective of the flashing frequency. The results obtained using driving simulators indicate that the glare from excessive brightness causes discomfort to drivers and is not conducive to sign recognition. The optimum luminance interval of LED signs was observed to vary with the changes in the ambient brightness. It is recommended using signs with steady LED transflective displays under poor lighting conditions.
  • Simulation and Analysis of Lateral Stability of Large Coach
    GUO Ying-shi, LU Yu-ping, FU Rui, YANG Fei
    China Journal of Highway and Transport. 2018, 31(4): 156-164,230.
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    To provide a theoretical basis for monitoring the lateral stability of an operating coach, based on the critical value simulation results of the state parameters under extreme conditions, the study of the boundary conditions for the stability of an operating coach was conducted. Based on the nonlinear three degree-of-freedom vehicle model, the extended Kalman filter state estimator was established to estimate the sideslip angle and the yaw rate of the operating coach in real time, and Trucksim was used to verify that the estimate has a better consistency and status-following capability. Using MATLAB/Simulink to establish the nonlinear seven-freedom-degree vehicle model, the influence of different running state parameters on the boundary of the β-β phase plane stability area was analyzed. Based on the simulation data, the boundary conditions of the stable region were based on the vehicle speed, the front wheel rotation angle, and the road surface adhesion coefficient, combining the control variable determined by β-β through the state estimation model. Standard stability tests of continuous sine steering angle input for the operating coach were performed in Trucksim, by analyzing whether the curve trend of the control variable during the running of the passenger bus exceeds the boundary of the stable area; subsequently, the running status of the vehicle is determined. The results show that when the operating coach was driven at 60 km·h-1 with a low road surface adhesion coefficient(μ=0.3) and a high road surface adhesion coefficient (μ=0.85) under a small steering angle, the yaw rate had can well follow the driver's intention (the curve tendency of the steering wheel angle) with a smaller delay response, and the vehicle is in a steady state. At this point, the control variable curve is always in the stable region. While inputting at a large steering angle under the same conditions, the yaw rate had not been able to follow the driver's intentions well, When driving in approximately 3.5 s below the low adhesion coefficient, the steering wheel angle had returned to the normal position, but the yaw rate was still at the maximum value; further, with a large delay, the operating coach had a sharp sideslip. When driving on a high adhesion coefficient road, the operating coach would have a serious offset at 2.5 s and 6.2 s. The vehicle is in an unstable state, and the curve of the control variable at the corresponding time exceeds the stable boundary. Herein, we verified the accuracy of the criteria for the lateral stability of the passenger bus.
  • Impact of Visual Secondary Task on Young Drivers' Take-over Time in Automated Driving
    LU Guang-quan, ZHAO Peng-yun, WANG Zhao-jie, LIN Qing-feng
    China Journal of Highway and Transport. 2018, 31(4): 165-171.
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    This study aimed to investigate the take-over response characteristics of young drivers in conditional automated driving. It was focused on the influence of the different visual tasks (3×3 and 4×4 arrow tasks) and the take-over-request lead time (TORlt, TTC is 5 s and TTC is 7 s) on the take-over time. We designed automated driving take-over scenarios including the different visual secondary tasks and take-over request lead time based on a driving simulator, and conducted the driving simulation tests with 29 young participants. The effects of the different visual tasks and take-over-request lead time on the take-over time were studied using the two-factor variance analysis. The Pearson correlation test was used to analyze the correlation of different take-over times. The results show that the take-over time with secondary tasks increase significantly compared to that without secondary tasks, and different tasks have no significant effect on the take-over time. Further, no significant interaction occurred between the secondary tasks and take-over time with an obstacle in the front. The take-over time decreased significantly in the take-over scenarios with an obstacle in the front compared to that in scenarios without an obstacle. Additionally, different TORlts had no significant effect on the steering reaction time and the brake reaction time when the participants engaged in different secondary tasks. For the take-over scenarios with an obstacle in the front, the participants tended to apply the operation of combined brake and swerve, and the ratio of the brake only to the combined operation for different TORlts is the same. A strong correlation exists between the take-over reaction time and the brake reaction time (r=0.7).
  • Path Planning and Tracking for Lane Changing Based on Model Predictive Control
    JI Jie, TANG Zhi-rong, WU Ming-yang, FANG Jing-cheng
    China Journal of Highway and Transport. 2018, 31(4): 172-179.
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    To address the problems of path planning and tracking for intelligent vehicles in the process of lane changing, the trapezoidal acceleration method was first adopted to design a virtual desired trajectory for a lane-changing maneuver. Adaptation of the proposed path planning method depends on the constraints of the total transition times, transient lateral acceleration, and associated jerk; thus, the mathematical relationship among the key variables above was calculated and their influence on the path planning was analyzed by drawing a virtual desired trajectory under different traffic scenarios. Next, a linear discrete-time vehicle dynamic model was presented and constraints on the control inputs, state values, and road parameters, used to ensure the tracking performance and vehicle stability, were encoded into the cost function; a model predictive controller with multi-constraints (MMPC) was then developed to track the desired trajectory for lane changing, and Hildreth's quadratic programming procedure was adopted to compute the front steering angle of the vehicle for path tracking during the lane-changing process. Finally, the proposed MMPC and a model predictive controller with a single constraint (SMPC) were developed using MATLAB and Carsim software, and both controllers for a path tracking task were implemented and compared in co-simulations under total transition times of 3 s and 6 s. When the total transition time was 6 s, both MMPC and SMPC were able to provide a satisfactory tracking performance for a lane change; in addition, compared with SMPC, the proposed MMPC was capable of providing satisfactory tracking performance and improving the maneuverability on a low-friction coefficient road when the total transition time was 3 s, which motivated the effort to improve the active safety of intelligent vehicles during the path tracking process.
  • Lane Change Merging Control Method for Unmanned Vehicle Under V2V Cooperative Environment
    ZHANG Rong-hui, YOU Feng, CHU Xin-nan, GUO Lie, HE Zhao-cheng, WANG Rong-ben
    China Journal of Highway and Transport. 2018, 31(4): 180-191.
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    Collaborative driving for the state transition of a vehicle platoon has become a popular research topic in the field of intelligent vehicle highway systems. It can effectively simplify the complexity of road traffic control management, reduce environmental pollution, and ensure road traffic safety. Based on a multi-intelligent vehicle cooperative driving control structure, this paper proposes a driving strategy for intelligent vehicles merging into a platoon, and analyzes the stable conditions for the driving state within a platoon of vehicles. After establishing the collaboration criteria and a safety assessment for when an unmanned vehicle merges into a vehicle platoon, based on high-order polynomials, the combined vehicle characteristics and effective motion trajectory for the lane changing of an unmanned vehicle are designed and obtained by introducing the index function of riding comfort. To assure vehicle safety and stability for the merging process of an unmanned vehicle, the motion relationship of each vehicle in the merging scenario was explored, including the movement relation between the merging vehicle and the front and rear vehicles of the platoon. The types and impact factors for a collision are analyzed, and the conditions for collision avoidance are provided. Based on the vehicle kinematics, a vehicle position error model is established, and the line speed and angular velocity are selected as inputs, combining the conditions of the system's large and progressive stability, as well as the path tracking controllers for the unmanned vehicle platoon merging process designed using the Lyapunov theory and Backstepping algorithm. Simulation and vehicle experiments show that the trajectories designed for merging into a vehicle platoon are feasible and safe. In addition, the controller demonstrates a better tracking performance. The longitudinal and lateral errors are within 15 cm, and the relative error of the direction deviation is less than 10%. This provides a useful reference for the unmanned state transition of a vehicle platoon and collaborative driving in an intelligent vehicle highway system, and can assist with the design and evaluation of road traffic safety in the future.
  • Asymptotically Sampling-based Algorithm with Applications to Autonomous Urban Driving on Structured Road
    SHAN Yun-xiao, GUO Xiao-min, LONG Jiang-yun, CAI Bin-bin, LI Bi-jun
    China Journal of Highway and Transport. 2018, 31(4): 192-201.
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    To solve the optimization and convergence problems during urban driving on the structured road, this study adopts an asymptotically optimal sampling-based framework for designing a planning algorithm applied to the autonomous vehicle. To improve the real-time computational problem, we chose an asymptotically optimal sampling-based method, called Stable Sparse RRT (SST), as the algorithmic frame, which does not require a time-consuming steering function to avoid solving the BVP. Furthermore, the "anytime" strategy was used to alleviate the time-consumption problem and improve the utilization efficiency of the optimal solution. In the algorithm, an improved close-loop control strategy was integrated to reduce the error between the planner and the real motion. In the closed loop system, a four-dimensional dynamic model was implemented to confirm the feasibility of the planned trajectory for the vehicle movement. We also designed an objective function to consider four weighted cost terms to ensure safety and comfort and tuned the weight coefficients under the driving conditions. Finally, we used an autonomous vehicle to conduct a field test along an open urban road as specified for the testing of autonomous vehicles. The testing scenario included avoiding a front static obstacle, following and overtaking a front dynamic obstacle, and applying a complex scenario with both dynamic and static obstacles. During the test, the velocity and steering data were collected to show the convergence properties and smoothness of the motion. The results demonstrate that the proposed algorithm can perform well when driving at a maximum of 30 km·h-1 during an obstacle-avoidance task, as well as with the following and overtaking tests. The vehicle can maintain the lane at 30 km·h-1, and avoid obstacles at a maximum speed of 15 km·h-1. When following the front car, the autonomous vehicle can accommodate a velocity required to maintain a reasonable distance and smooth driving.
  • Improving Traffic Safety via Traffic Flow Optimization of Connected and Automated Vehicles
    QIN Yan-yan, WANG Hao
    China Journal of Highway and Transport. 2018, 31(4): 202-210.
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    To improve the traffic flow safety of regular driven vehicles in terms of the rear-end collision, a stability optimal control method for the mixed traffic flow consisting of connected and automated vehicles (CAVs) and regular vehicles is proposed. The car-following model of the CAV was built using multiple vehicle feedbacks integrating speeds and accelerations, based on the full velocity difference model. Considering the random spatial distribution of vehicles in the CAV mixed traffic flow, the stability of various types of local platoons was considered as the optimization objective. The speed disturbance of the head vehicle of a local platoon was considered as the system input, while the speed disturbance of the tail vehicle was considered to be the system output. The stability constraints of local platoons were derived using the transfer function method of a classical control theory. In addition, stability charts with respect to the equilibrium speeds and CAV feedback coefficients were analyzed. Subsequently, the outputs of the CAV feedback coefficients were optimized, with each type of the local platoons being stable for any equilibrium speed as the control target. Numerical simulations were designed at the bottleneck of the freeway with an on-ramp. Subsequently, the impacts of optimal control for the CAV mixed traffic flow on vehicle rear-end collision risks were analyzed, under various conditions such as different CAV proportions. Our results show that the optimal control for the CAV mixed traffic flow can decrease vehicle rear-end collision risks. For the case of traffic flow with 100% CAV, the rear-end collision risks of regular vehicles flow can be reduced by more than 85.81% when the time-to-collision threshold is less than 2 s, while it can be decreased by 48.22%-78.80% if the time-to-collision threshold is more than 2 s. The optimal control method presented herein can effectively reduce the complexity of optimal control for large-scale CAV platoons. The research results can provide direct and theoretical references for the mixed traffic flow optimal control and strategies for improving vehicle rear-end collision safety under the background of a large-scale CAV flow.
  • A Blocking Efficiency Evaluation Method for Chains of Road Traffic Incident in IOV Environment（in English）
    LIANG Jun, ZHANG Wan-wan, HE Mei-ling, CHEN Long, XIONG Xiao-xia, CAI Tao
    China Journal of Highway and Transport. 2018, 31(4): 211-221.
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    In IOV(Internet of Vehicles)environment,in view of the efficiency of chains of road traffic incident (Chains of Road Traffic Incident, CRTI) blocking system, the reliability of blocking technique, the stability of blocking process and the economics of blocking cost were used as indicators of blocking strategy analysis, the first proposed a dynamic and real-time CRTI blocking efficiency evaluation algorithm in IOV environment. Risk sources are the root cause of the accident, the risk degree is determined by the risk sources, and it can show the possibility and severity of the accident, so the risk sources and the risk degree were taken as the evaluation indexes,risk sources were divided into two categories,the "driver-car-road-environment" risk degree and comprehensive evaluation function model were established. The risk sources could be obtained and screened in real-time for judging whether it needs to be blocked, then the optimal blocking strategy and blocking node was find, and the risk sources and risk degree were calculated before and after blocking,then the value range of the evaluation function was obtained through the substitution of the value of risk sources and risk degree. Finally, the comprehensive evaluation of blocking efficiency was achieved by analyzing the value comparison of the risk degree and the range of the cost function. The results show that the block system make the risk degree below 3 by comparing the risk degree before and after blocking, and ensure the traffic safety, overall, the driving risk is reduced by about 40%, so that the vehicle is in a safe state, and the success rate of blocking the CRTI is over 90%, shows that the blocking system can block the occurrence of CRTI; Through the comprehensive evaluation of the evaluation function, the cost value is basically between 0-1, indicating that the blocking efficiency of the blocking system is higher, and the comprehensive evaluation of CRTI blocking efficiency is achieved qualitatively and quantitatively.
  • Warning Effectiveness of Vehicle-to-infrastructure Cooperative Crossing Collision Prevention System at Non-signal Controlled Intersection （in English）
    LIU Kai, JIA Jie, LIU Chao, ANDO R
    China Journal of Highway and Transport. 2018, 31(4): 222-230.
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    To provide a better understanding of the warning effectiveness of the cooperative V2I crossing collision prevention system (CCPS), especially in understanding the various reactions of drivers with different mentalities and habits, a large-scale field of test (FOT) of the CCPS was conducted at a non-signal controlled intersection in Toyota city, Japan, through which the variation in the collision risks when the left/right turn vehicles attempted to cross an intersection and enter the primary road after receiving a CCPS information were studied. The FOT volunteer drivers' social attributes, transportation environments, the CCPS information types (including no information, providing both early warning information and emergency alarm, and providing only emergency alarm) were considered the influential factors in modeling the drivers' risk avoidance behavior in practice. The impacts of eight kinds of driving styles on the warning effectiveness of the CCPS, as well as the difference between the elderly and non-elderly drivers, were analyzed and compared. The Prais-Winsten regression model with panel-corrected standard errors and AR(1) was employed to solve the sequence correlation issue that root in repeat trials by volunteer drivers, which helped better understanding of the contribution of driving styles on the choices of time to collision. The results revealed that the CCPS is significantly effective for drivers with six kinds of driving styles, except for those with confidence in driving and with signal-oriented preliminary driving preparation. Further, different CCPS information types have different warning effectiveness on the elderly and non-elderly drivers. The current CCPS setting may increase the crash risk for elderly drivers with either the scrupulous driving tendency or the unstable driving tendency, as well as for the non-elderly drivers with the style of driving as a status symbol. The feasibility and reliability of the CCPS can be improved in the future through a personal-driving-style-oriented system design.
  • Investigation of Prediction Effectiveness of Brain Injury Criteria on Injury Risk Through Reconstruction of Pedestrian Impact Accidents
    WANG Fang, CAI Jia-xin, WANG Bing-yu, HAN Yong, LI Gui-bing, LI Fan
    China Journal of Highway and Transport. 2018, 31(4): 231-240.
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    To comprehensively explore the effectiveness of the various brain injury criteria, by the prediction and evaluation of pedestrian brain injuries involved in road traffic accidents, 10 real-world car-to-pedestrian impact accident cases that could be virtually reconstructed with detailed brain injury data were selected from a vehicle traffic accident database. The kinematics of the vehicles and pedestrians as well as the pedestrian brain injuries involved in the selected accident cases were constructed by means of the multibody and finite element methods, respectively. The correlations of several existing kinematics-based brain injury criteria with the pedestrian brain injuries observed in the accident cases were determined through the LASSO regression analysis method. The relationship between various strain-based brain injury criteria and two major types of brain injury (diffuse axonal injury DAI and brain contusions) was analyzed using injury risk curves recorded in the literature. The results indicate that the impact-power-based criterion, i.e., head impact power (HIP), which considers both the head translational and rotational motions, has a better correlation with the pedestrian brain injuries than the other kinematics-based injury criteria. The results also tend to suggest that the cumulative strain damage measure (CSDM) criterion is a better predictor for the DAI than the maximum principal strain (MPS) criterion; further, dilatation damage measure (DDM) presents a very low risk for brain contusions, which means that the DDM could not accurately predict brain contusions.
  • Reconstruction of Vehicle Movement in Video Sequences Based on One-dimensional Direct Linear Transformation（in English）
    FENG Hao, PAN Shao-you, HENG Wei-wei, ZHANG Ze-feng
    China Journal of Highway and Transport. 2018, 31(4): 241-249.
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    To overcome the problem that previous calculation methods could not reconstruct vehicle movement comprehensively and their application conditions were significantly limited, a new calculation method based on the one-dimensional Direct Linear Transformation(DLT), combined with the contour structure of the vehicle, was proposed to calculate the vehicle movement state in video sequences. The calibration information in the method was obtained from the contour structure of the target vehicle, which was not influenced by the calibration conditions of the road and environment, and this allowed expansion of its application. Meanwhile, the entire movement process of the vehicle in the video was fully encompassed by the calibration areas; therefore, the spatial wholeness of vehicle's movement track reconstruction was ensured. The calculations of the distance, speed, and acceleration were mutually independent, which eliminated cumulative errors. Finally, the confirmatory experiment of the vehicle driving in low speed, medium-high speed, and deceleration phase, as well as camera perspective directions of 90° and 30°, were conducted. The vehicle movements were calculated by this method, and then analyzed and compared with the real movement state recorded in the test. The results show that, in the camera perspective directions of 90° and 30°, respectively, regardless of the driving state, the errors between the calculated speed and the recorded speed were less than 1.5% and 4%, errors of distances were less than 3% and 5%, and errors of accelerations were less than 7% and 9%. This proves that the precision accuracies of both the speed and distance calculated by this method are improved greatly, and the accuracies of the calculated acceleration also satisfy the requirements. The results also demonstrate that the method is effective and practical in reconstructing vehicle movement in video sequences.
  • Parametric Research of Tire Mark Intensity in Accident Scene Based on Vehicle-road System
    MA Bin, XU Hong-guo, CHEN Yong, LIN Mu-yi
    China Journal of Highway and Transport. 2018, 31(4): 250-261.
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    To define the relationship between the intensity of tire marks and vehicle dynamics, and between the tire rubber wear mechanism and road gray feature, a parameter method to analyze the intensity of tire marks in accident scenes involving tires and vehicles are proposed. In consideration of the dynamic sliding friction model and tire dynamic feature, an improved nine-DOF vehicle-road nonlinear system model is established. Subsequently, the effectiveness of the model is evaluated through a real vehicle test using the vehicle stability detection system, which is based on the VBOX. The effects of the vehicle, tire, and road to the gliding friction force and tread wear feature are analyzed with the utilization of the tread wear energy model. Finally, a parameter method to analyze the intensity of tire marks is proposed based on the tire marks intension model. The simulation test is conducted under different braking forces, steering angles, three road types, and tire rubber to study the tire-road contact feature, tread wear amount, and visible tire marks feature. The results show that:the intensity of tire marks has a tight relation with gliding force regardless of the sliding direction; the key factors affecting the visible tire marks are the brake force and the road surface gray feature, and the steering angle has an evident effect judging by the visible tire marks irregular feature; the visible tire marks on the front tires appeared earlier, and have longer and higher intensions compared to the rear tires. The evidence shows that, the error to determine the vehicle speed based on the tire marks' starting point is inevitable and specific analyses should be performed according to the actual situation. The study provides a useful insight into accident rebuilding based on visible tire marks.
  • Projection Kinematics of Pedestrian Impacted by Minivans
    ZHANG Shi-bo, LIU Lan
    China Journal of Highway and Transport. 2018, 31(4): 262-269.
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    To enrich the kinematics theory of vehicle-pedestrian accidents, and provide support for technical analysis of minivan-pedestrian accidents, the post-impact kinematics of a pedestrian impacted by a minivan at an impact speed of 20-110 km·h-1 were investigated. PC-Crash, a multi-body-based accident modeling system, was used to construct simulation models of collisions between minivans and pedestrians; in addition, simulation tests using various groups of impact parameters were also carried out to produce post-impact kinematics data on pedestrians including the longitudinal/lateral throw distance, projection height, launch angle, rotation number, velocity of the pedestrian torso, and head acceleration. Reliable data on fourteen real-world accidents from the national automobile accident investigation system (NAIS) database were selected to verify the simulation results. arch projection was originally proposed and defined to describe the projection of pedestrians impacted by minivans in order to distinguish from wrap projection of pedestrians impacted by long-fronted vehicles, and forward projection of pedestrians impacted by plane-nosed vehicles. The results show that an impacted pedestrian will instantly be accelerated at up to 111%-127% of the impact speed of the vehicle in arch projection. The value of the head acceleration will exceed 3 000 m·s-2, and the value of the head injury criterion (HIC) will exceed 7 500 in high-speed impacts (110 km·h-1). The number of rotations in air does not exceed 3.0, the projection height does not exceed 4.0 m, and the launch angle is between 6° and 11°. The relationship between the pedestrian throw distance and the vehicle impact speed can be described using a power formula. In addition, the contact position, profile parameter of the vehicle front, pedestrian's walking speed, and impact posture have a certain extent effect on the pedestrian's projection kinematics (except for fender vault), and the extent of such influences is generally less than 5% (the maximum not exceeding 10%) in a typical arch projection. The corresponding speed for the safety threshold of a head injury (the value of HIC is 1 000) is approximately 55 km·h-1. The projection kinematics of fender vault has a large difference with arch projection, and the longest lateral throw distance of fender vault can reach up to 12.0 m.
  • Analysis of Urban Road Traffic Accidents Based on Improved K-means Algorithm（in English）
    GUO Lin, ZHOU Ji-biao, DONG Sheng, ZHANG Shui-chao
    China Journal of Highway and Transport. 2018, 31(4): 270-279.
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    To collect the data of urban road traffic accidents effectively and analyze the time-space characteristics and the cause of traffic accidents, data of 37654 accidents in the district of Yinzhou, Ningbo in the fourth quarter of 2016, were collected by a smart mobile application (APP). In view of the drawback of the traditional K-means clustering algorithm, i.e., slow convergence and low accuracy, an improved K-means clustering algorithm was proposed to eliminate the influence of outliers on the clustering results and identify the accident black spots automatically. The results show that the effective accident data sampled based on the mobile phone APP account for 96.4% of the actual alarm amount, which can meet the requirement of the accuracy and quality for the accident data analysis; the change trend of each quarter over the last four years' accidents exhibits an obvious erratic change. The number of accidents between motor vehicles was the highest, followed by the accidents involving motor vehicles and non-motorized vehicles, and the proportions were 58.4% and 15.8% respectively. In terms of temporal characteristics, the number of accidents on a Monday is the highest with the rate of 15.4%, the lowest on a Thursday with the rate of 13.2%; in terms of space characteristics, the locations of road traffic accidents are primarily concentrated in the ground sections, road intersections, and parking lot, where the accident rates are 77.4%, 11.6%, and 7.0% respectively. The accidents that occurred in the residential area and the elevated road are relatively low, with the rates of 3.2% and 0.9% respectively. In terms of the cause of accidents, driving behaviors such as following the front car closely, turning a corner without avoiding straight vehicles, illegal lane changing, and speeding are the primary reasons for accidents between two motor vehicles, with the rates of 28.8%, 22.9%,15.6%, and 7.6% respectively. Turning a corner without avoiding the oncoming non-motor vehicles, running the red light of non-motor vehicles, occupying the wrong lanes, and reverse driving of non-motor vehicles are the primary reasons for accidents between motor vehicles and non-motor vehicles, with the rates of 36.6%, 16.6%,9.9%, and 7.3% respectively.
  • Road Crash Risk Prediction Model for Continuous Streaming Data Environment
    GAO Zhen, GAO Yi, YU Rong-jie, HUANG Zhi-qiang, WANG Xue-song
    China Journal of Highway and Transport. 2018, 31(4): 280-287.
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    This paper describes research on a road crash risk prediction model for a continuous observation and dynamic management environment (called a continuous data environment) in an active traffic management (ATM) system. A traffic crash is an event with a small probability, and the ratio of crashes to non-crash cases in crash risk prediction research is not coordinated, and therefore poses the issue of an imbalanced data classification. To build a crash risk prediction model, existing research has been mostly based on a "matched case-control" under-sampling method to extract non-crash cases from continuous traffic flow data at a certain proportion; thus, the prediction accuracy of the model in a continuous data environment is inadequate. The research proposes utilizing a full set of traffic flow data to build a model and avoid an imbalanced data classification by "adjusting the classification threshold to discriminate crashes from non-crashes." The loop detector data and crash history data of the Shanghai expressway system from May to June 2014 were used experimentally. The area under an ROC curve (AUC) was used as an index to compare the commonly used crash risk prediction model (using logistic regression and random forest algorithms) based on the full set of data and the sample data respectively. The influence of three different classification thresholds (Youden's index, the crash occupancy, and the cross point method) on the comprehensive prediction accuracy of a crash and non-crash was compared using the geometric mean of sensitivity and specificity as the indices. The results show that, in a continuous data environment, the model with a full set of data improves the overall prediction capability by 13.06%. Youden's index method for the classification threshold calculation increases the optimal comprehensive prediction accuracy of crash and non-crash cases.
  • Safety Evaluation Method Based on Traffic Simulation and Extreme Value Theory
    WANG Chen, XIA Jing-xin, LU Zhen-bo, QIAN Zhen-dong
    China Journal of Highway and Transport. 2018, 31(4): 288-295,303.
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    To improve the performance of safety evaluation methods in China, this paper puts forward a new safety evaluation method for urban intersections by integrating microscopic traffic simulation and the extreme value theory (EVT). First, a new systematic simulation calibration method was proposed. The post encroachment time (PET) distribution was used as the calibration objective, constrained with the average queue length as a measure of effectiveness (MOE) and total conflict counts. The genetic algorithm (GA) was applied to find the optimal values of driver behavior models in simulation packages. Then, a peak-over-threshold (POT) based EVT was employed to predict the annual crash frequency of intersections, using the PET as input. Ten intersections in Fengxian, Shanghai were chosen for a case study. Three calibration methods were used to develop the simulation models, including a traditional method considering the queue length, a two-stage method considering both the MOE and conflict counts, and the proposed multi-constrained method for optimizing the PET distribution. Crash predictions were derived based on POT-based EVT and the three methods. The results show that the combination of POT and the proposed calibration method outperform the other two methods in terms of correlation, Spearman rank correlation, and crash prediction. Moreover, both field-based POT and simulation-based POT achieve a good crash prediction performance for crossing and rear-end conflict types, and an acceptable performance for a lane change conflict type. Thus, the proposed method can be considered a feasible and promising safety evaluation method for urban intersections.
  • Bayesian Traffic Conflict Model Accounting for Heterogeneity
    GUO Yan-yong, LIU Pan, WU Yao, LI Xiao-wei
    China Journal of Highway and Transport. 2018, 31(4): 296-303.
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    To develop a traffic conflict model at signalized intersections, traffic conflict data were extracted from 260 h of video data at 29 signalized intersections. Accounting for the traffic conflict heterogeneity, the random parameters Poisson-lognormal (RP-PLN) traffic conflict model and the random effects Poisson-lognormal (RE-PLN) traffic conflict model were developed. The posterior distributions of the models' parameters were estimated by the Bayesian method and the Markov chain Monte Carlo (MCMC) simulation. Using the deviance information criterion and the residual standardization decision coefficient, the goodness-of-fit of the models were compared. The results show that both the traffic conflict models can handle the traffic conflict heterogeneity; however, the goodness-of-fit of the RP-PLN traffic conflict model performs the RE-PLN traffic conflict model. Given that the other variables remained unchanged, a 1% increase in the crossing-through volume could increase the right-turn & crossing-through (RC) traffic conflict by 0.54%; a 1% increase in the right-turn volume could increase the RC traffic conflict by 0.65%; a 1% increase in the proportion of large vehicles could increase the RC traffic conflict by 1.06%; the raised and pavement channelized islands could decrease the RC traffic conflict by 18.9% and 17.3%, respectively; the acceleration lane could decrease the RC traffic conflict by 22.9%; a 1% increase in the right-turn radius could decrease the RC traffic conflict by 10.5%; the installation of a right-turn yield sign could decrease the RC traffic conflict by 16.5%; the setting up of the protected right-turn phase could decrease the RC traffic conflict by 29.8%.
  • Traffic Safety of Rainy Day Expressway with “Moving Dam”
    QIAO Jian-gang, WU Yan-xia, ZHANG Xue-jie
    China Journal of Highway and Transport. 2018, 31(4): 304-311.
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    To ensure the safety of expressway traffic and reduce the hidden dangers in terms of traffic safety under rainy conditions, the distance of the vehicle and the safety speed on an expressway during a rainy day were explored according to the actual traffic conditions and road surface runoff characteristics and their interaction. The concept of a "moving dam" was put forward for the first time, and the causes and formation mechanism of a "moving dam" phenomenon were analyzed. The relationship between the water film thickness and the critical velocity of hydroplaning during a "moving dam" was obtained through an analysis of the tire force. Using Fluent to simulate the obstruction of the vehicle to water, the critical distance of the outer lane vehicle at different speeds was determined according to the formation conditions of a continuous "moving dam". By using the principle of fluid mechanics, the relationship between the vehicle speed and water film thickness, and the hydrodynamic pressure of the tire, were analyzed to determine the critical velocity of hydroplaning of the inner lane car under different rainfall intensities. The results indicate that, under rainy conditions, the vehicles on the outer lane of expressway have a blocking effect on the road runoff, resulting in a "moving dam". Unlike a normal drainage state, under the action of a "moving dam", the thickness of the water film increases and the critical velocity of the hydroplaning of the inside lane car decreases. Under the set conditions, when the distance between the vehicles is less than 40 m, the dams of the two vehicles are continuous, and an increase in the time headway can weaken the role of a continuous "moving dam". During the course of the vehicle, the hydrodynamic pressure of the tire increases with an increase in the thickness of the water film and the operating speed of the cars. The probability of cars sliding over water under the action of a "moving dam" increases. According to the hydrodynamic pressure of the tire and the value of the water slide, the critical velocity of the hydroplaning corresponding to different rainfall intensities was determined, which can be used as a reference speed limit for vehicles under rainy conditions.
  • Traffic Assignment Method for Multiclass Users Considering Safety Reliability
    HUANG He-lai, JIANG Meng-xi, HAN Chun-yang, XU Guang-ming, WONG S C, ABDEL-ATY M
    China Journal of Highway and Transport. 2018, 31(4): 312-321.
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    To explore the influence of travel safety on users' travel choice behavior, a traffic assignment model for multiclass users was proposed considering the crash risk cost and travel time. According to the randomness of the accident characteristic, the path safety reliability was defined, and the safety preference of the users was reflected by calculating crash risk cost budget herein. Based on the basic crash prediction model associated with crash exposure and crash risk, the crash risk cost distribution of road links and intersections were defined considering the different characteristics of links and intersections. Further, the difference in the turn crash risk cost was contained in crash risk distribution model of intersections. A path-based solution algorithm and a successive average method were designed to solve the traffic assignment model, and the validities of the model and algorithm were analyzed by the numerical example. The results show that the safety reliability has significant influence on the users' travel choices. Considering the crash risk cost in the generalized travel cost, users are more likely to choose the paths with lower crash risk cost. Specifically, users with different risk tendencies have different choice characteristics. Conservative users tend to choose the paths with smaller standard deviations in crash risk cost, namely the paths with lower fluctuations in crash risk cost. Meanwhile, neutral users prefer to choose the paths with lower mean crash risk costs. Moreover, considering the crash risk cost of intersections will directly affect the traffic assignment results. Namely, the number of intersections in a path and the difference in turn will also affect the users' choice. Additionally, the model also shows the potential application value for passenger-volume forecasting, as well as network security assessment and management.
  • Game-theory-based Hazardous Materials Transport Network Routing
    LU Jian, LIU Yu-jie, MA Xiao-li
    China Journal of Highway and Transport. 2018, 31(4): 322-329.
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    On the condition that the traffic accident probabilities of transport road sections are unknown, a route-selection model is proposed by considering both transport cost and transport risk, aiming at reducing the impact of hazardous materials (HAZMATs) transport accidents on the nearby residents and surrounding environment. By considering the second accident caused by toxic gas leakage after a traffic accident, the accident consequence was quantified and defined as the affected number of people of diffusion coverage. In consideration of the complexity of roads in actual transport, a shortest-path algorithm was applied to select several alternative routes; therefore, the transport route network was simplified. Assuming that the accident was caused by an attacker and the transporter could use the combination of several routes, a compound route selection model based on the zero-sum game theory was built to minimize transport risk. The heuristic algorithm was applied to solve this model, thus the transport probabilities of each route chosen by the transporter could be obtained, which formed the compound route selection strategy. Considering liquid chlorine transportation in the Shanghai road network as a typical case, the compound route selection model was tested and verified. The strategy obtained from the model was then compared with the optimal single route selection strategy; the expected number of affected people of the former was 184, while that of the latter was 374. The results show that the compound route selection model can well solve the problem of unknown accident probabilities of HAZMAT transport road sections. The compound route selection strategy can well guide an actual HAZMAT transport and is better than the single route selection strategy for its much lower transport risk. The application of the shortest-path algorithm can well limit the transport route length and helps to save the transportation cost. The heuristic algorithm has a short running time; if the number of alternative routes is limited advisably, it can be used for much larger road networks.
  • Transportation Path Optimization for Hazardous Materials Considering Characteristics of Risk Distribution（in English）
    DAI Cun-jie, LI Yin-zhen, MA Chang-xi, CHAI Huo, LIU Lin-zhong
    China Journal of Highway and Transport. 2018, 31(4): 330-342.
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    To control the magnitude of transportation risk of hazardous materials (HAZMAT), and improve the rationality of a risk distribution in HAZMAT routing, and considering the interest in HAZMAT carriers, a multi-path combination optimization problem for multiple types of HAZMAT transported on the same road network was studied. First, a method for calculating the physical dissimilarity between paths was proposed based on the physical characteristics of the path, such as the risk value on each link, and a calculation method for the spatial dissimilarity between paths was presented according to the geospatial location of each link along a road network. The threshold of total risk and the minimum dissimilarity between paths were set as constraints to establish a multi-objective optimization model that considers the cumulative risk, transport cost, and transport time. An improved NSGA-Ⅱ was designed to solve this model in which the dynamic crowding distance was used to improve the uniformity of the individual distribution in the population. Finally, a random grid network was applied as an example to simulate and optimize the choice of transportation paths for a variety types of HAZMAT transported between different origin-destination nodes with single and multi-trip scenarios, and the actual data from a road transportation network were applied to verify the feasibility of the optimization method. The results indicate that setting the dissimilarity constraint between HAZMAT transportation paths can mitigate the number of shared links/nodes and avoid an excessive concentration of transportation risk in local areas; modestly increasing the total risk can improve the physical and spatial dissimilarity between paths, and further disperse the risk distribution, as well as expand the optimization space for the transportation paths of HAZMAT carriers. This study provides a new approach for government authorities in terms of risk control and risk distribution management and provides decision support for HAZMAT carriers with regard to transportation path selection.
News More>>
Download More>>
Links More>>
New Media
  • Wechat Account
  • WeChat video Account
Visited
    Total Visits:    Today:    Online:
Copyright © China Journal of Highway and Transport, All Rights Reserved.
Tel: 029-82334387　
E-mail：zgglxb@163.com
Powered by Beijing Magtech Co. Ltd,.