Driver behaviour is one of the most important factors recognised as a cause of road crashes. A driving simulator has become the key tool to simulate and verify the interaction between vehicle and driver in a well-conditioned traffic environment. This research aims at exploring driver behaviour in various scenarios by studying the relationship between physical, physiological, and attentional driving data. It also aims to discover the appropriate LOD (Level of Detail) of those scenarios that result in more reliability in a driving simulator. The study, developed in this thesis, is based on the data collected by the driving simulator and by the eye tracker devices of the i.Drive laboratory of Politecnico di Milano. The driving simulator is made up of a station with a fixed seat, monitors, a steering wheel, and pedals to drive the vehicle. Driving scenarios offer pathways characterized by different environmental details. By using the driver simulator, it is possible to observe the driver's behaviour as their LOD varies but unchanging the road infrastructure. The experiment uses four levels of detail from simple scenarios (having carriageway and road signs) to an entirely close to realistic scenario (with realistic buildings and all other elements like trees, lamps, pedestrians). The data analysis aims to search for the relationships between the representative data of vehicle (speed, break, gas pedal use and trajectory), the physiological data of the driver (skin conductance) and those attributable to the attention of the driver (fixation duration, pupil diameter of eye, and gaze positions). The general hypothesis of the research is “the level of detail of road scenarios for driving simulator affects driving behaviour” and “Relationship between perception signals and reaction signals depends on the time needed to take a decision” The relations between three types of signals are studied for some parameters such as the loop, the LOD, the type of road segment (straight or curved) and the type of drivers (identified by the clustering).
Il comportamento del guidatore è uno tra i principali fattori riconosciuti come causa di incidenti stradali. I simulatori di guida sono diventati una tecnologia fondamentale per simulare e quantificare l’interazione tra veicolo e guidatore in studiate condizioni di traffico. Questa tesi ha l’obiettivo di studiare il comportamento del guidatore in differenti scenari di simulazione analizzando la relazione tra i dati veicolari, fisiologici e attenzionali della guida. Inoltre, vuole definire quale sia il Livello di Dettaglio (LOD) degli scenari stradali che consente di conseguire la massima verosimiglianza di guida. Lo studio, che è stato sviluppato in questo lavoro di tesi, si basa sui dati raccolti dal simulatore di guida e dai dati del rilevatore oculare (eye tracker) del laboratorio i.Drive. Il simulatore di guida è composto da una postazione con sedile fisso, monitor, volante e pedaliere per la guida del veicolo. Gli scenari di guida propongono percorsi caratterizzati da differenti dettagli ambientali. Utilizzando il simulatore di guida, è possibile osservare il comportamento del conducente al variare del livello di dettaglio con cui è rappresentato lo scenario stradale. L’esperimento ha considerato quattro livelli di dettaglio (LOD): dalla sola sede stradale a uno scenario completamente realistico con resa foto realistica delle facciate degli edifici visibili dalla sede stradale. L’analisi dei dati si pone come obiettivo la ricerca delle relazioni tra i dati rappresentativi del veicolo (velocità, traiettoria), i dati fisiologici del guidatore (conduttanza cutanea) e quelli riconducibili all’attenzione del guidatore (fissazione, diametro oculare, posizione dello sguardo). L'ipotesi generale della ricerca è che “il LOD possa influenzare il comportamento alla guida” e “la relazione tra percezione e reazione alla guida dipenda dal tempo necessario a prendere una decisione”. Le relazioni tra tre tipi di segnale, dati fisici, fisiologici e di attenzione vengono studiate al variare di alcuni parametri come il numero di giro, LOD, tipo di segmento stradale (rettilineo o curvilineo) e tipo di guidatore (cluster di appartenenza).
The study of relationship between physical, physiological and attentional driving data in simulated road scenarios
Rawal, Bhuwan
2020/2021
Abstract
Driver behaviour is one of the most important factors recognised as a cause of road crashes. A driving simulator has become the key tool to simulate and verify the interaction between vehicle and driver in a well-conditioned traffic environment. This research aims at exploring driver behaviour in various scenarios by studying the relationship between physical, physiological, and attentional driving data. It also aims to discover the appropriate LOD (Level of Detail) of those scenarios that result in more reliability in a driving simulator. The study, developed in this thesis, is based on the data collected by the driving simulator and by the eye tracker devices of the i.Drive laboratory of Politecnico di Milano. The driving simulator is made up of a station with a fixed seat, monitors, a steering wheel, and pedals to drive the vehicle. Driving scenarios offer pathways characterized by different environmental details. By using the driver simulator, it is possible to observe the driver's behaviour as their LOD varies but unchanging the road infrastructure. The experiment uses four levels of detail from simple scenarios (having carriageway and road signs) to an entirely close to realistic scenario (with realistic buildings and all other elements like trees, lamps, pedestrians). The data analysis aims to search for the relationships between the representative data of vehicle (speed, break, gas pedal use and trajectory), the physiological data of the driver (skin conductance) and those attributable to the attention of the driver (fixation duration, pupil diameter of eye, and gaze positions). The general hypothesis of the research is “the level of detail of road scenarios for driving simulator affects driving behaviour” and “Relationship between perception signals and reaction signals depends on the time needed to take a decision” The relations between three types of signals are studied for some parameters such as the loop, the LOD, the type of road segment (straight or curved) and the type of drivers (identified by the clustering).| File | Dimensione | Formato | |
|---|---|---|---|
|
2021_06_RAWAL.pdf
accessibile in internet solo dagli utenti autorizzati
Descrizione: Thesis main file
Dimensione
29.36 MB
Formato
Adobe PDF
|
29.36 MB | Adobe PDF | Visualizza/Apri |
|
2021_06_RAWAL_Appendix.pdf
accessibile in internet solo dagli utenti autorizzati
Descrizione: Appendix file
Dimensione
29.23 MB
Formato
Adobe PDF
|
29.23 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/175309