Naslov
Development of methodology to calibrate a pedestrian microsimulation model

Autori
Gruden, Chiara

Vrsta, podvrsta i kategorija rada
Ocjenski radovi, doktorska disertacija

Fakultet
Fakultet za građevinarstvo, prometno inženjerstvo i arhitekturu

Mjesto
Maribor

Datum
29.09

Godina
2022

Stranica
192

Mentor
Šraml, Matjaž

Neposredni voditelj
Ištoka Otković, Irena

Ključne riječi
pedestrian ; microsimulation model ; calibration ; neural network ; surrogate safety indicators ; reaction time.

Sažetak
Walking, as a mode of transport, is becoming more and more widespread, in a world, where urban conglomerates are broadening and becoming denser and denser. Also, modern lifestyle trends on one side, and eco-friendly policies on the other, push people into walking habits, increasing the need for a suitable, attractive, accessible, connected, and safe walking infrastructure. To reach such a result, it is first all necessary to understand, what are the needs of the users of this infrastructure: in other words, we should take into consideration the behavioral specificities and the safety needs of pedestrians. In this process microsimulation models of pedestrian behavior, surrogate safety techniques, and technologies able to measure specific traits of pedestrian dynamics play a central role. The firsts allow reproducing repeatedly in a virtual environment a specific infrastructure and to study the response of pedestrians without the costs and drawbacks of the concrete realization of the infrastructure itself. Nevertheless, to be accurate and efficient, they need to go through long and tedious calibration and validation processes, which are often seen as important limitations by technicians. Surrogate safety techniques are methods, that saw their beginnings in the 60s, and that are based on the concept, that it is possible to predict the safety level of a location, using near-accidents, i.e. events in traffic, that have many similarities with crashes, but that – thanks to various local conditions – do not end in collisions. The main advantage of such techniques is that they are proactive, solving also the ethical issue of reactive techniques (such as accident data analysis), which wait for accidents to happen. Till this moment, these proactive techniques have been mainly applied to on-field measurements and are primarily centered on motorized road users, for whom also software to calculate surrogate safety measures starting from microsimulation results has been developed. Less interest has been shown for vulnerable road users, especially pedestrians, who have been less extensively studied and for whom - due also to their specific walking characteristics - still do not exist a similar program. Finally, an element that could highly affect pedestrian safety is the reaction time of these road users. As a matter of fact, pedestrians are the only road users, who can have a sudden reaction while involved in a conflict, which can turn the outcome of a conflict from a tragic end to only a bad memory. Nevertheless, it has been a big issue how to measure reaction time. Eye-tracking technology could be one of the solutions, allowing us to analyze the directions and objects fixated by pedestrians. These listed issues are also the topics that are addressed by this research work. Focusing on the study of the action of pedestrians while crossing the road on an unsignalized crosswalk set on a roundabout entry leg, the dissertation thesis aims at studying the crossing time, reaction time, and surrogate safety aspects typical of pedestrians at the recalled location. The main purpose of the research work is to develop a methodology to calibrate pedestrian Social Force Model (Helbing, 1995) at a selected location, using a specifically formulated neural network as a tool to fine-tune the model's behavioral parameters. To this aim, a literature review has been worked out and eight parameters have been chosen to be fine- tuned, five of whose are related to pedestrian behavior and three of them are related to car- following behavior. After the selection of input parameters, a feedforward network has been formulated, which also recognized those parameters as influential on the selected output to be measured (on-field), simulated (by the microsimulation model), and predicted (by the neural network), i.e. pedestrian crossing time. The first results of the neural network demonstrated its ability to predict crossing time and confirmed the opportunity to use it in the calibration process. Its further application in the framework of the whole calibration process has then brought considerably positive results, finding a combination of input parameters that improved the performance of the microsimulation model by 37 % in comparison to the default one, leading to an error between the micro-simulated crossing time and the on-field measured value of crossing time to 7.43 %. After two additional validation steps, the outputs of the calibrated model have been used to calculate three measures of surrogate safety, i.e. time-to- collision, time advantage, and relative speed, which are – according to the underwent literature review – the most applied and validated indicators for pedestrians. Also, in this case, results demonstrated an improvement in the calculation of surrogate safety measures when using the calibrated outcomes in comparison to their calculation of the “default” model outputs. Finally, reaction time measurement and prediction have been addressed by the thesis, in order to be able to describe pedestrian crossing action in its completeness. Quantitative eye-tracking outputs, i.e. fixations and their characteristics, have been the starting point for the calculation of pedestrian reaction time at different locations, and they allowed to create of a database of behavioral, geometric, regulatory, and flow characteristics, which was the foundation for the training and formulation of a new prediction model for pedestrian reaction time. The prediction model, which consists of a cascade- correlation neural network, gave a good response to the learning and generalization steps, turning a 74 % correlation between the measured reaction time values and the predicted ones, and being able to follow the variability of these values. Though overall promising results have been achieved, some limitations of the study should be taken into account and should be also considered as possible starting points for further research works. Focusing on the calibration methodology, at this point, it has not been automized. The full automatization of the method is one of the goals that have been set for future research works, and that can also facilitate the application of the same methodology to other locations. As a matter of fact, the use of the proposed calibration process to new infrastructure typologies would on a side, additionally validate the method, on the other side, it would provide sets of calibrated parameters related to each specific type of infrastructure, that could be of practical use for technicians. Also, the surrogate safety aspect could – and should – be improved. The thesis demonstrated that it is possible to calculate surrogate safety indicators for pedestrians starting from microsimulation outputs and that the calibrated results improved also the calculation of these safety measures. Nevertheless, more accurate outcomes are still needed. Finally, the measurement and prediction of reaction time could also be further developed, considering a broader sample of the pedestrian population, which could include various, more or less vulnerable pedestrian groups. This could allow us to discover problematic aspects for these specific groups, leading to solutions for even more safe and more accessible walking infrastructure.

Izvorni jezik
Engleski

Znanstvena područja
Građevinarstvo, Tehnologija prometa i transport

POVEZANOST RADA