The car's automation technology is progressing fast and has already reached a very good level of safety. Because of the advantages that autonomous cars will introduce, the question of how a driver wants to be driven becomes more and more important to ensure driving comfort for the passive driver and, so, fast and wide acceptance of this technology. The I.DRIVE (Interaction between Driver, Road Infrastructure, Vehicle and Environment) Lab of Politecnico di Milano, where this research was conducted, aims at developing inter-disciplinary proficiency required for the analysis and modeling of behavioral aspects due to the interaction between driver, vehicle, infrastructure, and environment. In this research thesis, we have developed a software and hardware platform that allows studying, in a real car, the interaction between driver, car, road infrastructure and environment. We focused on driver stress and the driving factors that impact driver stress to increase the little available knowledge in this field and help the acceptance of car automation in the future. We instrumented a vehicle with different sensors to acquire vehicle's and environmental's data. To assess the driver's stress with a continuous and objective measure, we acquired the driver's physiological signals; we developed software able to acquire the sensor's signals, and store them in a database. Furthermore, our software analyzes the vehicle and environmental data to extract information about the surroundings and stress indexes from the physiological data. Finally, we used the data in a correlation analysis using the K-nn algorithm to evaluate the correlation between stress indexes and driving features. To validate our framework, six drivers performed a manual drive, an autonomous drive, and a manual drive as passengers. We validated our platform with the data acquired in these experiments. Moreover, we evaluate the correlation using our software pipeline. Results demonstrate that longitudinal jerk, the angular velocity on x-axis, and the linear acceleration on y-axis have a high correlation with the skin conductance's phasic component and the respiration rate that are two stress indexes. The ratio between low frequency and high frequency and the heart rate of the electrocardiogram showed a low correlation with all driving features studied.
La tecnologia di automazione delle auto sta progredendo rapidamente e ha già raggiunto un ottimo livello di sicurezza. Visto i vantaggi che le auto autonome garantiranno, la questione di come un guidatore vuole che la macchina guidi diventa sempre più importante per garantire il comfort di guida per il guidatore passivo e, quindi, una rapida e ampia accettazione di questa tecnologia. Il Laboratorio I.DRIVE (Interaction between Driver, Road Infrastructure, Vehicle and Environment) del Politecnico di Milano, dove è stata condotta questa ricerca, mira a sviluppare le competenze interdisciplinari richieste per l'analisi e la modellizzazione degli aspetti comportamentali dovuti all'interazione tra conducente, veicolo, infrastruttura e ambiente. In questa tesi di ricerca abbiamo sviluppato una piattaforma software e hardware che permette di studiare, in un'auto reale, l'interazione tra guidatore, automobile, infrastruttura stradale e ambiente. Ci siamo concentrati sullo stress del guidatore e sui fattori di guida che influiscono sullo stress del guidatore per ampliare la scarsa conoscenza disponibile in questo campo e aiutare l'accettazione dell'automazione automobilistica in futuro. Abbiamo installato sul veicolo diversi sensori per acquisire i dati del veicolo e ambientali. Per valutare lo stress del guidatore con una misura continua e oggettiva, abbiamo acquisito i segnali fisiologici del guidatore; abbiamo sviluppato un software in grado di acquisire questi segnali e memorizzarli in un database. Inoltre, il nostro software analizza i dati del veicolo e ambientali per estrarre informazioni sull'ambiente circostante e indici di stress dai dati fisiologici. Infine, abbiamo utilizzato i dati ottenuti per un'analisi di correlazione utilizzando l'algoritmo K-nn per valutare la correlazione tra gli indici di stress e i parametri di guida. Per convalidare il nostro framework, sei conducenti hanno eseguito una guida manuale, una guida autonoma e una guida manuale come passeggeri. Abbiamo validato la nostra piattaforma utilizzando i dati acquisiti in questi esperimenti. Inoltre, abbiamo valutato la correlazione tra questi dati utilizzando la nostra pipeline software. I risultati dimostrano che il jerk longitudinale, la velocità angolare sull'asse x e l'accelerazione lineare sull' asse y hanno un'alta correlazione con la componente fasica della conduttanza della pelle e la frequenza respiratoria che sono due indici di stress. Il rapporto tra bassa frequenza e alta frequenza e la frequenza cardiaca dell'elettrocardiogramma ha mostrato una bassa correlazione con tutte le caratteristiche di guida studiate.
Studying user acceptance of autonomous driving in the wild
Gabrielli, Alessandro
2020/2021
Abstract
The car's automation technology is progressing fast and has already reached a very good level of safety. Because of the advantages that autonomous cars will introduce, the question of how a driver wants to be driven becomes more and more important to ensure driving comfort for the passive driver and, so, fast and wide acceptance of this technology. The I.DRIVE (Interaction between Driver, Road Infrastructure, Vehicle and Environment) Lab of Politecnico di Milano, where this research was conducted, aims at developing inter-disciplinary proficiency required for the analysis and modeling of behavioral aspects due to the interaction between driver, vehicle, infrastructure, and environment. In this research thesis, we have developed a software and hardware platform that allows studying, in a real car, the interaction between driver, car, road infrastructure and environment. We focused on driver stress and the driving factors that impact driver stress to increase the little available knowledge in this field and help the acceptance of car automation in the future. We instrumented a vehicle with different sensors to acquire vehicle's and environmental's data. To assess the driver's stress with a continuous and objective measure, we acquired the driver's physiological signals; we developed software able to acquire the sensor's signals, and store them in a database. Furthermore, our software analyzes the vehicle and environmental data to extract information about the surroundings and stress indexes from the physiological data. Finally, we used the data in a correlation analysis using the K-nn algorithm to evaluate the correlation between stress indexes and driving features. To validate our framework, six drivers performed a manual drive, an autonomous drive, and a manual drive as passengers. We validated our platform with the data acquired in these experiments. Moreover, we evaluate the correlation using our software pipeline. Results demonstrate that longitudinal jerk, the angular velocity on x-axis, and the linear acceleration on y-axis have a high correlation with the skin conductance's phasic component and the respiration rate that are two stress indexes. The ratio between low frequency and high frequency and the heart rate of the electrocardiogram showed a low correlation with all driving features studied.| File | Dimensione | Formato | |
|---|---|---|---|
|
Studying user acceptance of autonomous driving in the wild.pdf
accessibile in internet per tutti
Dimensione
27.67 MB
Formato
Adobe PDF
|
27.67 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/10589/170871