With progress in enabling autonomous cars to drive safely on the road, a new concern is emerging about how they should be driving. A common thought is that they should adopt their users’ driving style, however, new studies suggest that this is not the best approach. The goal of this work is to assess the environmental causes of the driver’s stress using both environmental and physiological data coming from sensors. In particular, we used a modified car on which we installed cameras and Inertial Measurement Unit (IMU), magnetometer, Global Positioning System (GPS) and Light Detection and Ranging (LiDAR) sensors. We considered data coming from two different acquisitions: in the former, there was just one participant that drove the car manually; in the latter, there were six participants that experienced three driving conditions: manual, autonomous and passenger. In both acquisitions we measured the Electrocardiogram, the Skin Conductance and the pupil diameter of the driver and, in the second acquisition, we measured also the respiration. Once collected the data, a software system processed them and provided information about how environmental variables affect the level of stress of the driver. Results show that there are some environmental variables that, depending on the level of stress of the driver, can assume values with a higher probability. In particular, speed, time headway, longitudinal acceleration, yaw and the steering angle are the variables that change the most. Results also show that the three driving conditions of the second acquisition have different models of stress and confirm literature results on how the physiological data change in stressful situations.
Grazie ai progressi che permettono alle autovetture autonome di guidare in sicurezza sulle strade, un nuovo quesito sta sorgendo a proposito dello stile di guida che dovrebbero avere. Un pensiero comune è che esse debbano adottare lo stile di guida del conducente, ma recenti studi hanno dimostrato che questo approccio non è il migliore. L’obiettivo di questa tesi è di valutare le cause ambientali del livello di stress del conducente utilizzando i dati ambientali e fisiologici provenienti dai sensori. In particolare, abbiamo utilizzato una autovettura modificata, sulla quale sono state installate diverse telecamere e alcuni sensori quali Inertial Measurement Unit (IMU), magnetometro, Global Positioning System (GPS) e Light Detection and Ranging (LiDAR). Abbiamo considerato dati provenienti da due acquisizioni differenti: nella prima, vi era un solo partecipante che guidava la macchina manualmente; nella seconda, vi erano sei partecipanti che hanno provato tre condizioni di guida: manuale, autonoma e come passeggero. In entrambe le acquisizioni abbiamo misurato l’Elettrocardiogramma, la conduttanza della pelle e il diametro della pupilla del conducente e, nella seconda acquisizione, abbiamo misurato anche la respirazione. Una volta collezionati i dati, un sistema software li ha processati fornendo informazioni su come le variabili ambientali influenzino il livello di stress del conducente. I risultati mostrano che vi sono delle variabili ambientali che, a seconda del livello di stress del conducente, possono assumere determinati valori con una maggiore probabilità. In particolare, la velocità, il time headway, l’accelerazione longitudinale, l’imbardata e l’angolo di sterzata sono le variabili che cambiano di più. I risultati mostrano anche che le tre condizioni di guida della seconda acquisizione hanno modelli di stress differenti e confermano i risultati della letteratura su come i dati fisiologici cambino in situazioni stressanti.
I.DRIVE : assessing environmental causes of driving stress through Bayesian networks
Mattioli, Daniele
2019/2020
Abstract
With progress in enabling autonomous cars to drive safely on the road, a new concern is emerging about how they should be driving. A common thought is that they should adopt their users’ driving style, however, new studies suggest that this is not the best approach. The goal of this work is to assess the environmental causes of the driver’s stress using both environmental and physiological data coming from sensors. In particular, we used a modified car on which we installed cameras and Inertial Measurement Unit (IMU), magnetometer, Global Positioning System (GPS) and Light Detection and Ranging (LiDAR) sensors. We considered data coming from two different acquisitions: in the former, there was just one participant that drove the car manually; in the latter, there were six participants that experienced three driving conditions: manual, autonomous and passenger. In both acquisitions we measured the Electrocardiogram, the Skin Conductance and the pupil diameter of the driver and, in the second acquisition, we measured also the respiration. Once collected the data, a software system processed them and provided information about how environmental variables affect the level of stress of the driver. Results show that there are some environmental variables that, depending on the level of stress of the driver, can assume values with a higher probability. In particular, speed, time headway, longitudinal acceleration, yaw and the steering angle are the variables that change the most. Results also show that the three driving conditions of the second acquisition have different models of stress and confirm literature results on how the physiological data change in stressful situations.| File | Dimensione | Formato | |
|---|---|---|---|
|
Thesis.pdf
solo utenti autorizzati dal 08/04/2022
Dimensione
7.18 MB
Formato
Adobe PDF
|
7.18 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/10589/175555