No description available.

No description available.

No description available.

Mémoire numérisé par la Division de la gestion de documents et des archives de l'Université de Montréal.

Mémoire numérisé par la Division de la gestion de documents et des archives de l'Université de Montréal

Ces travaux de thèse portent sur la mise en place d’une stratégie de sonification qui vise à proposer un retour sonore pouvant se substituer au bruit moteur dans les véhicules électriques, rendant au conducteur les informations qu’il transmet habituellement sur la dynamique du véhicule.Pour cela, nous nous sommes basés sur une première phase d’analyse qui nous a permis d’étudier comment le bruit automobile influence notre perception du mouvement. A partir de deux expériences menées en simulateur de conduite, nous avons pu relier le retour sonore et la vitesse perçue par le conducteur, définissant ainsi la métaphore du bruit moteur sur laquelle s’appuie le contrôle des sons de synthèse.De façon similaire au bruit moteur, le retour sonore proposé informe le conducteur par l’intermédiaire de sa variation de hauteur tonale. Pour arriver à un son qui informe le conducteur efficacement et qui reste acceptable sur toute la gamme de vitesse du véhicule, nous nous sommes appuyés sur l’illusion de glissando infini de Shepard-Risset, qui nous permet de donner un retour d’information précis grâce à une hauteur tonale qui varie rapidement tout en restant contenue dans une plage de fréquence restreinte.L’apport de cette stratégie a enfin été testé lors de deux expériences, la première portant sur l’influence de ce retour sonore sur la perception de la vitesse des conducteurs, la seconde sur leur comportement dans une tâche de freinage. Ces deux études ont montré un effet significatif du retour sonore qui suggère que ces informations sont bien intégrées par les conducteurs, faisant de ces sons un candidat prometteur pour devenir le « bruit moteur » des véhicules de demain. / This thesis aims to build an auditory display to sonify electric vehicles. Our goal consisted in bringing back to the driver the motion information, which is usually provided by the combustion engine noise.The first stage of this work consisted in analyzing how automotive noises can influence drivers’ perception of motion. We conducted two driving simulator experiments to study drivers’ speed perception in presence of different automotive noises. These results provided a link between the acoustic feedback and the speed perceived by the driver, on which we based our sonification strategy.Similarly to combustion engine noise, the acoustic feedback proposed in this work informs the driver via its pitch variation. We used the Shepard Risset glissando illusion to sonify the whole speed range of the vehicle. Pitch circularity in the construction of these sounds provides a precise information on small speed variation with fast pitch variations, and is in addition restrained within a narrow bandwith.We then tested the contribution of this strategy in two experiments. The first dealt with the influence of the proposed sounds on drivers’ speed perception ; the second with their behavior in a common braking task. These studies showed that the drivers easily integrate the information brought by this sound, and that it influences their perception of motion and modifies their driving behavior. These inputs make the proposed sound a good candidate to become the new « engine noise » of future electric cars.

The objective of the current study is to evaluate the use of virtual reality (VR) headsets to measure driving performance. This is desirable because they are several orders of magnitude less expensive and, if validated, could greatly extend the powers of simulation. Out of several possible measures of performance that could be considered for evaluating VR headsets, the current study specifically examines drivers’ latent hazard anticipation behavior both because it has been linked to crashes and because it has been shown to be significantly poorer in young drivers compared to their experienced counterparts in traditional driving simulators and in open road studies. The total time middle-aged drivers spend glancing at a latent hazard and the average duration of each glance was also compared to these same times for younger drivers using a VR headset and fixed-based driving simulator. In a between-subject design, forty-eight participants were equally and randomly assigned to one out of four experimental conditions – two young driver cohorts (18 – 21 years) and two middle-aged driver cohorts (30 – 55 years) navigating either a fixed-based driving simulator or a VR-headset-based simulator. All participants navigated six unique scenarios while their eyes were continually tracked. The proportion of latent hazards anticipated by participants which constituted the primary dependent measure was found to be greater for middle-aged drivers than young drivers across both platforms. Results also indicate that the middle-aged participants glanced longer than their younger counterparts on both platforms at latent hazards, as measured by the total glance duration but had no difference when measured by the average glance duration. Moreover, the difference in the magnitude of performance between middle-aged and younger drivers was the same across the two platforms. There were also no significant differences found for the severity of simulator sickness symptoms across the two platforms. The study provides some justification for the use of virtual reality headsets as a way of understanding drivers’ hazard anticipation behavior.

As technology is becoming ever more present in the world, the use of simulators as a teaching tool is more prevalent than ever, with good evidence to back up their usefulness. Drivers’ education is one area among many that is seeing this change. These types of simulators are often classified by their level of fidelity, or how alike the simulator is to reality. However, fidelity is a far from precise term with many sub-categories that different researchers treat differently. This means, along with a lack of research, that the knowledge on how fidelity affects learning in simulations is diffuse. This study aimed to investigate how specifically changing the immersive feeling of a truck simulator affected the speed of learning to reverse with a virtual truck and trailer. More immersion was achieved by constructing a crude roof and walls on the simulator mimicking the inside of a vehicle. Results showed that there was no significant increase in the learning speed. However, there was a relatively near significant difference (p=0.148) between the groups, where more immersion was linked to faster learning. Participants also judged the simulator as being more useful in their learning if experiencing the immersive setup, which could be a potential cause of the near significant difference. Because of the low sample size of 23, no definitive conclusions could be drawn from this study. Taken together, it seems that a larger study stands a good chance of revealing the effect simulator fidelity can have on learning from a relatively simple change of immersion.

The Type-II dilemma zone refers to the segment of roadway approaching an intersection where drivers have difficulty deciding to stop or proceed through at the onset of the circular yellow (CY) indication. Signalized intersection safety can be improved when the dilemma zone is correctly identified and steps are taken to reduce the likelihood that vehicles are caught in it. This research employs driving simulation as a means to collect driver response data at the onset of the CY indication to better understand and describe the dilemma zone. The data obtained was compared against that from previous experiments documented in the literature and the evidence suggests that driving simulator data is valid for describing driver behavior under the given conditions. Fuzzy logic was proposed as a tool to model driver behavior in the dilemma zone, and three such models were developed to describe driver behavior as it relates to the speed and position of the vehicle. These models were shown to be consistent with previous research on this subject and were able to predict driver behavior with up to 90% accuracy. / Graduation date: 2013

Fahrerassistenzsysteme (FAS), wie zum Beispiel die „Aktive Gefahrenbremsung“, sollen dazu beitragen, das Fahren sicherer zu machen und die Anzahl an Unfällen und Verunglückten im Straßenverkehr weiter zu senken. Bei der Entwicklung von FAS muss neben der funktionalen Zuverlässigkeit des FAS sichergestellt werden, dass der Fahrer die Assistenzfunktion versteht und fehlerfrei benutzen kann. Zur Bestimmung geeigneter Systemauslegungen kommen in der Entwicklung Probandenversuche zum Einsatz, bei denen die zukünftigen Nutzer das FAS erleben und anschließend beurteilen. In dieser Arbeit wird die Eignung eines statischen Fahrsimulators für die Durchführung von Probandenversuchen zur Bewertung aktiv eingreifender FAS untersucht. Hierzu wurde ein Fahrversuch auf der Teststrecke und im statischen Fahrsimulator konzipiert, mit jeweils ca. 80 Probanden durchgeführt und die Ergebnisse bezüglich der Auswirkung des FAS „Aktive Gefahrenbremsung“ auf ausgewählte objektive und subjektive Kennwerte in der jeweiligen Versuchsumgebung vergleichend gegenübergestellt. Es zeigt sich, dass der statische Fahrsimulator prinzipiell für die Durchführung von Studien zur Bewertung aktiv eingreifender FAS geeignet ist. Als Ergebnis der Arbeit werden Erkenntnisse zur Aussagekraft der betrachteten Kennwerte sowie Empfehlungen zur Versuchsdurchführung im statischen Fahrsimulator gegeben.