Overview
ABSTRACT
Driver fatigue is still poorly detected, but is responsible for a large number of accidents. Current detection systems are based either on the driver's physical state, or on the state of the vehicle. Physiological data on the driver would help to detect fatigue, but this would be invasive and unreliable in cases of high sleep debt. What's more, fatigue can be perceived in very different ways from one individual to another, making its analysis all the more complex. This article takes a look at the data that can be acquired from both the driver's and the vehicle's point of view: the first is based on vehicle and driver characteristics, the second feeds an expert system.
Read this article from a comprehensive knowledge base, updated and supplemented with articles reviewed by scientific committees.
Read the articleAUTHORS
-
Alexandre LAMBERT: Doctoral student - ECE-Paris, Paris, France
-
Céline BARTH: Teacher-researcher - ECE-Paris Laboratory, Paris, France
-
Manolo HINA: Teacher-researcher - ECE-Paris Laboratory, Paris, France
-
Amar Ramdane CHERIF: Professor - University of Versailles Saint-Quentin, LISV Laboratory, Vélizy, France
-
Aakash SONI: Teacher-researcher - ECE-Paris Laboratory, Paris, France
-
Assia SOUKANE: Research Director - ECE-Paris, Paris, France
INTRODUCTION
Fatigue at the wheel is a phenomenon with many causes, resulting in accidents that can cause material and human damage. Everyone agrees on this, but when it comes to detecting fatigue at the wheel in order to prevent these accidents, things get complicated. Is there a direct way of measuring fatigue? Which sensors are the most suitable? What parameters and values should detection be based on? Is a detection system effective for people with different profiles? Many questions can arise from the issue of driver fatigue.
By drawing up a portrait of fatigue according to different communities, we will explore its manifestations, causes and ways of experiencing it. The many factors that make up and influence fatigue make it complex to model. Researchers use a variety of methods to acquire parameters, with or without contact. It is also possible to observe the driver's driving behavior and the environment in which he or she operates. The heterogeneity of the data requires intelligent systems to aggregate it and make decisions in real time.
A wide variety of intelligent systems exist. They are often broken down into three parts: perception organs, processing and reasoning organs, and decision organs. To detect fatigue at the wheel, perception can be provided by cameras or biosensors. Processing and reasoning are based on Machine Learning (ML) techniques. Other techniques, such as rule-based systems, are also used, but are less effective than Deep Learning (DL), a sub-family of techniques within machine learning, which is more flexible and enables models to better adapt to the complex data generated by driver fatigue.
However, we still have a long way to go before we can design systems that are reliable and truly detect driver fatigue. In the various stages of designing intelligent systems, it is possible to introduce biases in data collection, processing and reasoning. Finally, new fields of research in artificial intelligence, such as "explicability", are opening up new avenues of reflection for developing more robust models.
Exclusive to subscribers. 97% yet to be discovered!
Already subscribed? Log in!
KEYWORDS
intelligent systems | face detection | biosensors | vehicle behavior
Driver fatigue detection
Fatigue
Article included in this offer
"Vehicules and mobility"
(
90 articles
)
Updated and enriched with articles validated by our scientific committees
A set of exclusive tools to complement the resources
Bibliography
Exclusive to subscribers. 97% yet to be discovered!
Already subscribed? Log in!
