Making autonomous vehicles safer
Context and challenges
Next generation of autonomous cars (level 3 automation), are expected to come in the following years. In these vehicles, the driver does not have to monitor the road and can engage in secondary activities.
However, in case of unexpected incidents, the driver must take the control back when notified by the car. This situation is called a takeover, and the vehicle convey the need for takeover via a takeover request. The takeover request can take multiple forms and use different modalities, the more common being haptic, visual and auditory. These takeovers are dangerous situations that should be treated carefully.
Objectives
Our main objective for this project was to find how to support the driver in case of takeover.
- This means finding a definition for takeover quality, as well as metrics to quantify it.
- Using this information, a review of the literature would highlight the factors influencing this quality.
- These factors should be monitored in order to train Machine Learning models to predict takeover quality.
- Acting on these factors would then allow us to impact the takeover quality.
Partners and funding
HEIA-FR, Humantech Institute
University of Fribourg, DIUF
EPFL+ECAL Lab
Funded by Hasler Foundation
Results
A smart agent was designed, implemented and evaluated to recommend takeover request modalities in order to improve takeover quality. Along the way, several analysis and Machine Learning models were done and disseminated.
Valorisation
Capallera, M., Meteier, Q., de Salis, E., Angelini, L., Carrino, S., Khaled, O. A., & Mugellini, E. (2019, September). Owner manuals review and taxonomy of ADAS limitations in partially automated vehicles. In Proceedings of the 11th International Conference on Automotive User Interfaces and Interactive Vehicular Applications (pp. 156-164).
Meteier, Q., Capallera, M., Angelini, L., Mugellini, E., Khaled, O. A., Carrino, S., … & Boll, S. (2019, September). Workshop on explainable AI in automated driving: a user-centered interaction approach. In Proceedings of the 11th International Conference on Automotive User Interfaces and Interactive Vehicular Applications: Adjunct Proceedings (pp. 32-37).
Capallera, M., de Salis, E., Meteier, Q., Angelini, L., Carrino, S., Khaled, O. A., & Mugellini, E. (2019, September). Secondary task and situation awareness, a mobile application for conditionally automated vehicles. In Proceedings of the 11th International Conference on Automotive User Interfaces and Interactive Vehicular Applications: Adjunct Proceedings (pp. 86-92).
de Salis, E., Baumgartner, D. Y., & Carrino, S. (2019, September). Can we predict driver distraction without driver psychophysiological state? a feasibility study on noninvasive distraction detection in manual driving. In Proceedings of the 11th International Conference on Automotive User Interfaces and Interactive Vehicular Applications: Adjunct Proceedings (pp. 194-198).
de Salis, E., Capallera, M., Meteier, Q., Angelini, L., Abou Khaled, O., Mugellini, E., … & Carrino, S. (2020, July). Designing an AI-companion to support the driver in highly autonomous cars. In International Conference on Human-Computer Interaction (pp. 335-349). Springer, Cham.
Meteier, Q., Capallera, M., de Salis, E., Angelini, L., Abou Khaled, O., Mugellini, E., … & Carrino, S. (2020). Inside the cockpit of semi-autonomous cars of tomorrow. In Proceedings of GSGS’20: 5th Gamification & Serious Game Symposium, September-November 2020, Switzerland (No. CONFERENCE). September-November 2020.
de Salis, E., Meteier, Q., Pelletier, C., Capallera, M., Angelini, L., Sonderegger, A., … & Carrino, S. (2021, August). Clustering of Drivers’ State Before Takeover Situations Based on Physiological Features Using Unsupervised Machine Learning. In International Conference on Human Interaction and Emerging Technologies (pp. 550-555). Springer, Cham.
Meteier, Q., de Salis, E., Capallera, M., Widmer, M., Angelini, L., Abou Khaled, O., … & Mugellini, E. (2022). Relevant Physiological Indicators for Assessing Workload in Conditionally Automated Driving, Through Three-Class Classification and Regression. Frontiers in Computer Science, 3.
de Salis, E., Meteier, Q., Capallera, M., Pelletier, C., Angelini, L., … & Carrino, S. (2022). Predicting Takeover Quality in Conditionally Autonomous Vehicles based on Takeover Request Modalities, Driver Physiological State and the Environment. In Intelligent Human Systems Integration (IHSI 2022). AHFE International.
de Salis, E., Meteier, Q., Capallera, M., Pelletier, C., Angelini, L., … & Carrino, S. (2022). Machine Learning Agent to Recommend the Best Modality for Takeover during Conditionally Automated Driving In Human Interaction and Emerging Technologies (IHIET-AI 2022). AHFE International.
Meteier, Q., Capallera, M., De Salis, E., Widmer, M., Angelini, L., Abou Khaled, O., … & Sonderegger, A. (2022). Carrying a passenger and relaxation before driving: Classification of young drivers’ physiological activation. Physiological reports, 10(10), e15229.
