The goal of the project was the modelling and simulation of the interaction between human-driven and autonomous vehicles, based on field observations in which I used eye tracking glasses for the generation of the data.
Process
In order to achieve the most reliable and realistic possible model, I had to follow this procedure:
Field observation of the interaction between drivers on urban road. Experienced drivers were asked to drive their car in a predefined urban course while wearing an eye glass mounted gaze sensor.
I watched and analysed videos of the participants' eye gaze and scene video as well as his/her retrospective commentary, of a total 800mins.
I modelled the interaction between the drivers using the observation study. The model implemented in Matlab and was based on the game theory and the utility function to determine the cost of the driver in yielding in an interaction with other driver.
Urban traffic was simulated in VISSIM software paired with Matlab, to evaluate the credibility of the designed interaction model.
From the simulation of the driver interaction model, metrical results were extracted
I compared the model with the real study results, thus verifying the effectiveness and credibility of the model.
Tools used
Matlab
VISSIM and SUMO traffic simulators
Tobii eye-tracking glasses and the corresponding software
Remarks
Thesis project's grade: 10/10
Second award for my school's most innovative project in 2018
First Cypriot undergraduate students' project scholarship award, for my graduate thesis, in 2017 by Youth Board of Cyprus.