Reliability analysis of a driving simulator to reproduce vehicle dynamics from a microscopic point of view

• Autor(a) principal: Fernandes, Beatriz da Silva
• Data de Publicação: 2022
• Tipo de documento: Dissertação
• Idioma: eng
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: http://hdl.handle.net/10773/36206
• topic: Road transport; Driving simulator; Emission models; Reliability; Validation

Despite the constant technological innovation, the road transport sector remains a significant source of pollutant emissions. Research on solutions to increase the sustainability of vehicles and reduce their emissions is a timely topic. Although real-world experiments can be considered, these yield some drawbacks, like increased levels of pollution and risks of accidents. Thus, simulator-based experiments can be relevant to studying vehicle dynamics and driver behaviour. However, the scientific usefulness and robustness of the simulator are severely compromised if the reliability of the simulation results is not ensured. The present dissertation focuses on an exploratory analysis that aims to study the reliability of a fixed-based driving simulator to reproduce driving parameters for emission estimation by analyzing the acceleration and speed of 2 drivers in a simulation environment and in a real environment. Tests conducted on both a driving simulator and real-world cases were performed on an urban and a highway scenario. Vehicle dynamic parameters, such as instantaneous speed, acceleration, and position, were collected using the driving simulator software (Carnetsoft simulator) in the simulation tests, while an in-vehicle OBD (on-board diagnostics) device was used under the empirical tests. The data sample was divided into ten events that typically occur on urban and highway trips to better proceed with the comparative evaluation of the obtained results. The pollutant emissions were estimated using the Vehicle-Specific-Power (VSP) methodology. A comparative evaluation and a two-sample Kolmogorov-Smirnov test (K-S test) were conducted to validate the obtained results and assess the reliability of a driving simulator in reproducing vehicle dynamics and emissions. The results of this exploratory validation analysis showed that, with a relative error of 4% and accounting for the small number of participants, the total average emissions of all events were not significantly different (958,39 g for simulated and 998,06 g for empirical tests). The K-S test results revealed that the VSP mode distributions did not follow the same pattern in 4 out of 10 events, namely "left-turn after a traffic light", "entering highway", "exiting highway" and "moving forward" for driver 1 and "right-turn with priority", "give way", "exiting highway" and "moving forward" for driver 2. This means that the drivers displayed different behaviours in both the simulated and empirical tests for those specific events. It was found that in general, the driving simulator can replicate vehicle dynamics from a microscopic perspective. Nevertheless, a deep study on driver’s behaviour in a simulator compared to a real-world environment should be put in place for more rigorous results.