Do multimodal interfaces affect mode awareness in partially automated vehicles? - CNRS - Centre national de la recherche scientifique Accéder directement au contenu
Communication Dans Un Congrès Année : 2022

Do multimodal interfaces affect mode awareness in partially automated vehicles?

Résumé

In partially automated vehicles, the driver can delegate lateral and longitudinal control of the car to Advanced Driver-Assistance Systems (ADAS). These systems are able to perform the driving task in specific conditions, defined by the manufacturer. When conditions are no longer met, for example passing through a bend with an elevated speed, the automated lateral control of the car ceases functioning, giving back the control of the steering wheel to the driver, but keeping automated longitudinal control activated. This type of situation can cause mode confusion if the driver does not understand that the interface informs them that they need to control the trajectory of the vehicle and not the speed (Endsley, 2017). These surprising situations are to be avoided to minimize risks and uncomfortable moments for drivers and can be caused by an inadequate mode awareness. Mode awareness can be represented as a subpart of situational awareness, dedicated to the functioning of ADAS (Kurpiers et al., 2020) and is influenced by multiple factors. One of them is mental workload (Endsley, 2018). Most of today’s partially automated vehicle’s interfaces use visual resources to inform on the state of ADAS. Driving being mainly a visual task, performing the driving task and monitoring the state of ADAS are sharing the same attentional resource which can cause a cognitive overload, resulting in a degradation of mode awareness. Several resource channels can be used to perform a task (Wickens, 2008). In highly automated vehicles, take-over requests exploiting visual and auditory interfaces allowed to reduce take-over time (Zhang et al., 2019). In partially automated vehicles, using multiple resources to transmit information on the state of ADAS should increase mode awareness. In the present driving simulator-based study, we investigated the effect of the interface modality on mode awareness, trust in automation, usability of interfaces and mental workload in partially automated vehicles. The simulated situations were based on real situations in which ADAS deactivate and potentially cause a loss of mode awareness. Forty participants (M = 53.3; SD = 6.8) were randomly assigned to one of two interfaces: a visual only interface, or a multimodal interface, presenting information about limits of ADAS in peripheral vision, auditory and haptic signals. Participants drove in six scenarios containing events that caused deactivations of ADAS (i.e., bends, slow vehicles, erased road markings, fog). The quality of take-overs and the visuo-attentional behaviors were exploited as a measure of mode awareness (Janssen et al., 2019). After the driving periods, participants filled out questionnaires investigating their mental models regarding functioning of the system. They were interviewed and filled out mental workload, usability, and trust scales. Results of this study are still in the processing phase and will be presented during the congress. They will contribute to develop interfaces of partially automated vehicles in order to induce adequate mode awareness and therefore reduce risks and uncomfortable situations for drivers. It will also inform researchers on the effect of multimodal interfaces of partially automated vehicles on different dimensions such as trust, workload and usability.
Fichier non déposé

Dates et versions

hal-03724042 , version 1 (15-07-2022)

Identifiants

  • HAL Id : hal-03724042 , version 1

Citer

Noé Monsaingeon, Loïc Caroux, Sabine Langlois, Céline Lemercier. Do multimodal interfaces affect mode awareness in partially automated vehicles?. 13th International Conference on Applied Human Factors and Ergonomics - AHFE 2022, Jul 2022, New York, NY, United States. ⟨hal-03724042⟩
121 Consultations
0 Téléchargements

Partager

Gmail Facebook X LinkedIn More