The German highway network hast o face new challenges in the near future, e.g. increasing traffic density and loads, climate change effects and new quality requirements regarding sustainability. It is necessary to come up with foresighted concepts in the present to be prepared for these challenges. Therefore it is important to adapt and enhance innovative attempts, which take changing impacts into account. One goal of these efforts is the development of adaptive systems for the provision of information and a holistic evaluation in real time. The paper describes the recent research and developments on a system for information and holistic evaluation in real time, taking into account sensor networks, evaluation procedures and their implementation in existing maintenance and inspection strategies.
In the paper it is investigated to what extend one can extrapolate the detailed accident database GIDAS (German In-Depth Accident Study), with survey area Hanover and Dresden region, to accident behavior in other regions and countries within Europe and how such an extrapolation can be implemented and evaluated. Moreover, it is explored what extent of accident data for the target country is necessary for such an extrapolation and what can be done in situations with sparse and low accident information in a target region. It will be shown that a direct transfer of GIDAS injury outcomes to other regions does not lead to satisfactory results. But based on GIDAS and using statistical decision tree methods, an extrapolation methodology will be presented which allows for an adequate prediction of the distribution of injury severity in severe traffic accidents for European countries. The method consists essentially of a separation of accidents into well-described subgroups of accidents within which the accident severity distribution does not vary much over different regions. In contrast the distribution over the various subgroups of accidents typically is rather different between GIDAS and the target. For the separation into the subgroups meaningful accident parameters (like accident type, traffic environment, type of road etc.) have been selected. The developed methodology is applied to GIDAS data for the years 1999-2012 and is evaluated with police accident data for Sweden (2002 to 2012) and the United Kingdom (2004 to 2010). It is obtained that the extrapolation proposal has good to very good predictive power in the category of severe traffic accidents. Moreover, it is shown that iterative proportional fitting enables the developed extrapolation method to lead to a satisfactory extrapolation of accident outcomes even to target regions with sparse accident information. As an important potential application of the developed methodology the a priori extrapolation of effects of (future) safety systems, the operation of which can only be well assessed on the basis of very detailed GIDAS accident data, is presented. Based on the evaluation of the presented extrapolation method it will be shown that GIDAS very well represents severe accidents, i.e. accidents with at least one severely or fatally injured person involved, for other countries in Europe. The developed extrapolation method reaches its limits in cases for which only very little accident information is available for the target region.
The German highway network is facing new challenges in the near future. The structures have to deal with increasing traffic loads, climate change effects and new requirements regarding sustainability while they are getting older and budget cuts can be expected. To guarantee a reliable highway network, it will be vital to adapt and enhance innovative approaches. Current bridge management relies on the results of conventional bridge inspections and thus has certain limitations when it comes to insufficient load bearing capacity and other systematic weaknesses. Therefore, new approaches for real time condition assessment of critical road infrastructure elements are to be developed.
To assess occupant safety in a crash test, criteria associating the measurements made with a crash test dummy to injury risk are necessary. To enable better protection of elderly car occupants the objective of this study was to develop improved thoracic injury criteria for the THOR average male dummy. The development of these criteria is usually based on matched dummy and Post Mortem Human Surrogate (PMHS) tests by relating the obtained PMHS injuries to dummy measurements. This approach is limited, since only a few tests in relevant loading conditions are available and any new test series requires high efforts to be performed due to their complexity and costs. To overcome these limitations and to extend the dataset for the development of THOR dummy chest injury risk functions a simulation-based approach was applied within the EC funded project SENIORS (Safety Enhanced Innovations For older Road Users - www.seniors-project.eu). Within this study frontal impact sled simulations with an FE model representing a THOR average male dummy and matched simulations with a human body model (HBM) representing an elderly car occupant were carried out. The HBM used for this study was the THUMS TUC with modified rib cage, which was developed in SENIORS. The modifications included material and geometry changes aiming to represent an elderly car occupant. The rib fracture risk was predicted with a deterministic approach whereby a rib was considered broken when the strain exceeded an age-dependent threshold. Furthermore, a probabilistic method was applied to predict the probability of sustaining a certain number of fractured ribs by comparing local strain values to the distribution of cortical rib ultimate strain. By relating the output from the HBM simulations to a multi-point dummy injury criterion, injury risk curves were calculated by statistical methods. The wide range of loading conditions resulted in the desired range of injuries and THOR ATD output. The number of fractured ribs predicted by the HBM based on the deterministic prediction method was between 0 and 15. Furthermore, the probabilistic risk for the number of rib fractures equal or greater than two, three or four was calculated for each load case. The THOR rib deflection criterion Rmax was between 18 and 56 mm, while the PC Score was in the range of 2.5 to 7.2. Based on these outputs new risk curves for the predicted deterministic (AIS2+/3+) and probabilistic injury risk were calculated. The new curves show reasonable shapes and significance that provide trust in their application. The new risk curves are compared to risk curves obtained by traditional methods. The results were found similar to previous injury risk functions based on physical tests, which gives a high level of confidence in the chosen approach. The simulation-based approach of matched ATD model vs. HBM simulation was successfully applied. Rmax curves show a slightly better quality than the injury criterion PC Score.
At the end of each year, the German Federal Highway Research Institute (BASt) publishes the road safety balance of the closing year. They describe the development of accident and casualty numbers disaggregated by road user types, age groups, type of road and the consequences of the accidents. However, at the time of publishing, these series are only available for the first eight or nine months of the year. To make the balance for the whole year, the last three or four months are forecasted. The objective of this study was to improve the accuracy of these forecasts through structural time-series models that include effects of meteorological conditions. The results show that, compared to the earlier heuristic approach, root mean squared errors are reduced by up to 55% and only two out of the 27 different data series yield a modest rise of prediction errors. With the exception of four data series, prediction accuracies also clearly improve incorporating meteorological data in the analysis. We conclude that our approach provides a valid alternative to provide input to policy makers in Germany.
The main focus of the benefit estimation of advanced safety systems with a warning interface by simulation is on the driver. The driver is the only link between the algorithm of the safety system and the vehicle, which makes the setup of a driver model for such simulations very important. This paper describes an approach for the use of a statistical driver model in simulation. It also gives an outlook on further work on this topic. The build-up process of the model suffices with a distribution of reaction times and a distribution of reaction intensities. Both were combined in different scenarios for every driver. Each scenario has then a specific probability to occur. To use the statistical driver model, every accident scene has to be simulated with each driver scenario (combinations of reaction times and intensities). The results of the simulations are then combined regarding the probabilities to occur, which leads to an overall estimated benefit of the specific system. The model works with one or more equipped participants and delivers a range for the benefit of advanced safety systems with warning interfaces.
The study aimed at estimating the impact of pedelecs (with an assumed higher speed than bicycles) on the traffic accident severity in Germany for different penetration rates. The analysis shows that in many real situations (68%) an electrical support of bicycles has no influence on the sequence of accident events. Taking into account a number of unreported "single bicycle accidents", the adoption of similar traffic behavior and similar age distribution, the authors determined a shift of 400 former slightly to seriously injured cyclists in Germany per year. Overall this would be an increase of approximately 2.3% in case of 10% of pedelec penetration with the pessimistic assumption of 10 km/h speed increase although first natural driving studies predict a much lower average speed increase of pedelecs. The hypothesis verbalized in the initial question whether a higher distribution of pedelecs will result in more severe accidents in Germany is not verified. The study shows that electrical support didn"t result in higher collision speed in general. In many accident situations, the speed of pedelecs has only a minor influence on the accident severity. Further research focusing on a possible change of driver behavior especially in new target groups (elderly people) will be needed.
This study aimed at developing an injury estimation algorithm for AACN technologies for Germany and compared them to findings based on Japanese data. The data to build and to verify the algorithm was obtained from the German in-depth Accident Database (GIDAS) and split into a training and a validation dataset. Significant input variables and the generalized linear regression model to predict severe injuries (ISS>15) were selected to maximize area under the receiver operating characteristic curve (AUC). Probit regression with the input parameter multiple impact, delta v, seatbelt use and impact direction gave the largest AUC of 0.91. Sensitivity of the algorithm was validated at 90% and specificity at 76% for an injury risk threshold of 2%. It appears that no major differences between Japan and Germany exist for injury estimation based on delta v and impact direction. However, far side impact and multiple crash events appear to be associated with a larger risk increase in the German data.
The project UR:BAN "Cognitive assistance (KA)" aims at developing future assistance systems providing improved performance in complex city traffic. New state-of-the-art panoramic sensor technologies now allow comprehensive monitoring and evaluation of the vehicle environment. In order to improve protection of vulnerable road users such as pedestrians and cyclists, a particular objective of UR:BAN is the evaluation and prediction of their behaviour and actions. The objective of subproject "WER" is development support by providing quantitative estimates of traffic collisions at the very start and predict potential in terms of optimized accident avoidance and reduction of injury severity. For this purpose an integrated computer simulation toolkit is being devised based on real world accidents (GIDAS as well as video documented accidents), allowing the prediction of potential effectiveness and future benefit of assistance systems in this accident scenario. Subsequently, this toolkit may be used for optimizing the design of implemented assistance systems for improved effectiveness.
The evaluation of the expected benefit of active safety systems or even ideas of future systems is challenging because this has to be done prospectively. Beside acceptance, the predicted real-world benefit of active safety systems is one of the most important and interesting measures. Therefore, appropriate methods should be used that meet the requirements concerning representativeness, robustness and accuracy. The paper presents the development of a methodology for the assessment of current and future vehicle safety systems. The variety of systems requires several tools and methods and thus, a common tool box was created. This toolbox consists of different levels, regarding different aspects like data sources, scenarios, representativeness, measures like pre-crash-simulations, automated crash computation, single-case-analyses or driving simulator studies. Finally, the benefit of the system(s) is calculated, e.g. by using injury risk functions; giving the number of avoided/mitigated accidents, the reduction of injured or killed persons or the decrease of economic costs.