Vehicle Systems and Driver Modelling

DSP, human-to-vehicle interfaces, driver behavior, and safety

Huseyin Abut, John Hansen, Gerhard Schmidt, Kazuya Takeda, Hanseok Ko, Huseyin Abut, John Hansen, Gerhard Schmidt, Kazuya Takeda, Hanseok Ko

• Publisher: De Gruyter
• Year: 2017
• Print ISBN: 9781501512124
• eBook ISBN: 9781501504167
• Edition: 1
• Topic: Technology & Engineering
• Subtopic: Automotive Transportation & Engineering
• Index: Technology & Engineering

World-class experts from academia and industry assembled at the sixth Biennial Workshop on Digital Signal Processing (DSP) for In-Vehicle Systems at Korea University, Seoul, Korea in 2013. The Workshop covered a wide spectrum of automotive fields, including in-vehicle signal processing and cutting-edge studies on safety, driver behavior, infrastructure, in-vehicle technologies.

Contributors to this volume have expanded their contributions to the Workshop into full chapters with related works, methodology, experiments, and the analysis of the findings. Topics in this volume include:

• DSP technologies for in-vehicle systems

• 
  
  

• Driver status and behavior monitoring

• 
  
  

• In-Vehicle dialogue systems and human machine interfaces

• 
  
  

• In-vehicle video and applications for safety

• 
  
  

• Passive and active driver assistance technologies

• 
  
  

• Ideas and systems for autonomous driving

• 
  
  

• Transportation infrastructure

• 
  
  

Part I:Vehicle System and Safety

Navid Shokouhi, Seyed Omid Sadjadi, Amardeep Sathyanarayana, and John H.L. Hansen

1Analysis of in-vehicle speech activity towards driver safety assessment

Navid Shokouhi: Center for Robust Speech Systems (CRSS), Dept. Of Electrical Engineering,

Jonsson School of Engineering & Computer Science, The University of Texas at Dallas,

Richardson TX-75080, USA. e-mail: [email protected]

Seyed Omid Sadjadi: Center for Robust Speech Systems (CRSS), Dept. Of Electrical Engineering,

Jonsson School of Engineering & Computer Science, The University of Texas at Dallas,

Richardson TX-75080, USA. e-mail: [email protected]

Amardeep Sathyanarayana: Center for Robust Speech Systems (CRSS), UTDrive lab,

Dept. Of Electrical Engineering, Jonsson School of Engineering & Computer Science,

The University of Texas at Dallas, Richardson TX-75080, USA. e-mail: [email protected]

John H.L. Hansen: Center for Robust Speech Systems (CRSS), UTDrive lab, Dept. Of Electrical Engineering, Jonsson School of Engineering & Computer Science, The University of Texas at Dallas,

Richardson TX-75080, USA. e-mail: [email protected]

Abstract: In-vehicle conversations commonly take place either as a result of drivers speaking to passengers or on the phone. Many conversations are beneficial in keeping drivers alert, especially in long distance monotonous driving. On the contrary, there are also instances where a conversation may adversely influence driving performance. In this study, we present a framework for the analysis of in-vehicle speech activity as a potential stimulus for driver distraction. Here the focus is on competitive behaviors in conversations which is considered a key factor affecting driving performance. The amount of overlapped speech and frequency of turn takings in a conversation are used as metrics to indicate competitiveness. A series of speech-based detection systems are presented. The output of these detection systems are analyzed alongside driving performance. The goal of this study is to take advantage of audio and vehicle-related signals recorded through a smart portable tablet from inside the vehicle and to determine the effects of engaging in in-vehicular conversations as a potential cause for driver distraction. Results show a strong correlation between competitive in-vehicular speech activity and driver distraction.

Keywords: In-vehicle speech activity detection, overlapped speech, turn taking rate, driver safety assessment.

1.1Introduction

In-vehicle speech activity is typical in the form of drivers’ conversations with passengers or on the phone. Many of these conversations are beneficial in keeping the driver alert, especially in long distance driving. On the contrary, there are also instances where a conversation may adversely influence driving performance. In an argumentative conversation, the driver is more likely to be distracted which could result in life-threatening errors. In such conversations, the parties exhibit a competitive behavior in maintaining dominance in the discussion. Such “competitive” conversations are what the driver must avoid. A third type of conversation, which is a direct result of advancements in voice-based technologies, is the driver’s interaction with gadgets. Voice-based navigation systems and cellular devices are examples of such technologies. Due to commercial popularity, these gadgets become more interactive as technology advances and there should be a higher level of control to determine whether the driver is spending excessive attention on the device [1]. The 100-car Naturalistic Study reported that driver inattention and human error contributes to over 75 % of the crashes, and 65 % of near crashes [2], causing not only an economic impact but also human loss and suffering. There are several factors contributing to human error in driving scenarios, one of which is driver distraction. Many regulations in the form of new laws and guidelines are enforced on drivers as well as automobile manufacturers to minimize distractions within t...

Table of contents

• Cover
• Title Page
• Copyright
• Contents
• Preface
• List of contributing authors
• Part I: Vehicle System and Safety
• Part II: Driver Modeling
• Part III: Signal Processing for HVI
• Index