Human-Automation Interaction Design

Developing a Vehicle Automation Assistant

Jediah R. Clark, Neville A. Stanton, Kirsten Revell

• Publisher: CRC Press
• Year: 2021
• ISBN:
  9781000464597
• Edition: 1
• Topic: Technology & Engineering
• Subtopic: Automotive Transportation & Engineering
• Index: Technology & Engineering

This text presents a four-step approach for applying communicative concepts to driving automation, including: scoping, piloting, designing, and testing. It further provides experimental data on how practical human-human communication strategies can be applied to interaction in automated vehicles.

The book explores the role of communication and the nature of situation awareness in automated vehicles to ensure safe and usable automated vehicle operation. It covers the issue of interaction in automated vehicles by providing insight into communicative concepts, the transfer of control in human-teams, and how these concepts can be applied in automated vehicles. The theoretical framework is built on by presenting experimental findings, design workshop output and providing a demonstration of prototype generation for automated assistants that addresses a wide range of performance outcomes within human-machine interaction.

Aimed at professionals, graduate students, and academic researchers in the fields of ergonomics, automotive engineering, transportation engineering, and human factors, this text:

• 
  
  

• Discusses experimental findings on how practical human-human communication strategies can be applied to interaction in automated vehicles.

• 
  
  

• Provides a four-step approach for applying communicative concepts to driving automation, including: scoping, piloting, designing and testing.

• 
  
  

• Explores the role of distributed situation awareness in automated vehicles.

• 
  
  

• Covers communication and system awareness in response to multiple complex road scenarios.

• 
  
  

• Provides design guidelines for automation-human handover design.

1
Introduction

DOI: 10.1201/9781003213963-1

1.1 BACKGROUND

A driverless future where automated systems are able to control a road vehicle and make strategic decisions for the driver promises a wide range of benefits including a reduction in road traffic accidents, an increase in users’ free time, and an increase in traffic and fuel efficiency (DFT, 2015). This automated future is one that has been embraced by the UK government (DFT, 2015) and has marked the start of a race among manufacturers to roll out models with increasingly sophisticated automated features (Fagnant & Kockelman, 2015). Given the myriad of benefits and the optimism showed by many, a critical and skeptical approach has been advised by the UK government and independent institutes to ensure that safety standards are adhered to (GOV, 2017).

The levels of automation put forward by SAE international (SAE, 2016) attempt to tighten the discourse surrounding automated vehicles (AVs) and their operational capacity by categorizing automated features into six discrete levels outlined in Table 1.1.

---

TABLE 1.1 SAE Levels of Automation (SAE, 2016)

Level	Name	Control Executor	Monitoring Environment	Fallback Performance	Capability (modes)
0	No automation	Driver	Driver	Driver	n/a
1	Driver assistance	Driver and system	Driver	Driver	Some
2	Partial automation	System	Driver	Driver	Some
3	Conditional automation	System	System	Driver	Some
4	High automation	System	System	System	Some
5	Full automation	System	System	System	All

---

Source: Summary adapted from SAE (2016)

These levels outline vehicles with no automated features (level 0) up to vehicles that require no human driver inputs to operate effectively (level 5). Between these two extremes, four levels of automation represent varying degrees of human driver tasks and responsibilities—each requiring both driver and automation to perform certain tasks at certain intervals. These levels, due to issues surrounding shared control and responsibility, exhibit novel vulnerabilities. For example, incident reports cite driver distraction, overreliance, and human error as being the central cause of collisions in AVs that require monitoring (level 2 automation; Banks et al., 2018; BBC, 2020; Stanton et al., 2019; SAE, 2016). Despite this, increasingly ‘sophisticated’ AVs are being made available to the public with the driver being further removed from aspects of the driving task. Beta-test AVs are currently equipped with level 2 automation—lateral and longitudinal automation that requires the human driver to monitor the driving task in the case of an emergency. However, level 3 AVs such as Audi’s A8 model equipped with traffic jam assist are available to citizens of states where level 3 AVs are road-legal. Level 3 AVs require transitions in control and responsibility as a result of breaching an operational limit (such as loss of central reservation detection in the Audi A8; Audi, 2018; SAE, 2016). These vehicles differ from their level 2 predecessors in that they allow the driver to take part in non-driving-related tasks while automation is in control (SAE, 2016). Level 4, in extension, may feature control transfers although level 4 AVs are assumed to not require falling back to the driver in the case of a breach of operational safety.

The recurring feature of requiring control transitions in shared-control AVs has been identified as contributing toward novel vulnerabilities such as a reduction in situation awareness (Endsley, 1995; Sarter & Woods, 1992, 1995), deterioration in attentional resources (Young & Stanton, 2002a, 2002b), mode error (Norman, 2015; Sarter & Woods, 1995), deskilling (Bainbridge, 1983), and lack of calibration in trust (Koo et al., 2015; Lee & See, 2004). Other factors arising from an increase in driver–automation interaction include the acceptance and usability of the technology (NSAI, 2018; Nwiabu & Adeyanju, 2012; Ponsa et al., 2009; Schieben et al., 2011).

AV technology is progressing quickly, and research must keep up with public and manufacturing demands. As legality of level 3 AVs is granted across the world, issues introduced by shared control and responsibility must be addressed to reduce fatalities and collisions while ensuring that the technology benefits the user with regards to usability and acceptance. This book considers such implications and develops foundations and design solutions for human–machine interfaces (HMIs) to optimize these human factors outcomes in level 3 and 4 AVs as a collective.

1.2 RESEARCH MOTIVATION

Ensuring that developments in automated technology are implemented safely allows manufacturers and the public to benefit from positive outcomes while ensuring that novel vulnerabilities are minimized. By conducting research into its safe development, a driverless future is more likely to be beneficial to societies and becomes a technology that should not be feared but embraced. The content within this book was acquired as part of a broader research project, Human Interaction: Designing Autonomy in Vehicles (HI:DAVe), which in turn was part of a nationwide research program, Towards Autonomy: Smart and Connected Control, funded by both Jaguar Land Rover and the EPSRC. HI:DAVe addressed the human–machine interfaces (HMIs) in level 3 and 4 AVs, giving particular attention to the situations where a takeover request is made as a result of a critical or noncritical operational/design violation (e.g., upcoming geographical boundary; SAE, 2016).

HMIs are central to the solutions posited to reduce vulnerabilities in shared-control AVs as they allow for the driving system to relay information between both driver and vehicle. As driver and automation have distinct roles in the future of AVs, the importance of effective HMIs is ever increasing. To inform design, this book draws upon concepts that aim to improve interaction from theory (distributed situation awareness—DSA; Stanton et al., 2006 & joint activity (JA) framework; Bradshaw et al., 2009; Clark, 1996; Klein et al., 2004, 2005) and practice (shift handover in human teams; e.g., Kerr, 2002). This is with a view of generating a novel HMI design that relays important information to the driver to improve safety following transitions of control while maximizing usability and optimizing trust and workload. Further, this book illustrates an example design lifecycle for human factors practitioners to draw inspiration from, progressing as follows: scoping, piloting, designing, and finally prototype testing, with each chapter being a part of a step in this progression.

The book approaches the issue of ‘handover’ (defined in this book as the transition of control from vehicle to driver) in an innovative way. The majority of previous research in level 3 AV technology is primarily concerned with transitions of control in response to emergency situations. Transitions of control should be central in discussions on how to improve human–automation interaction in level 3 AVs; however, this book acknowledges that this is an oversimplification of the vulnerabilities in level 3 AVs as handovers may be initiated by either party in response to a variety of events (Mirnig et al., 2017). Further, knowledge on how transitions should occur during nonemergency scenarios is limited, and design solutions that attempt to optimize outcomes are yet to be provided. This book, therefore, provides a unique perspective on the AV handover task by integrating communicative concepts found in other domains to improve communication throughout the automated cycle while ensuring that communication can be made more efficient and tailored to the driver by applying concepts of distributed situation awareness. In doing so, this book contributes to the body of knowledge on how to design HMIs to improve human factors outcomes in AVs requiring control transitions. Improving communication for level 3 vehicles may be a priority; however, findings can translate to automated levels that require both driver and automation to fulfill specific roles in the driving task. Findings from this book can be readily applied to level 4 vehicles as both level 3 and 4 AVs have the potential for human and automation to transfer control and responsibility between one another. It is therefore hoped that the output provided here will inform future HMI design in AVs regardless of specific target context.

1.3 RESEARCH OUTCOMES AND HYPOTHESES

This book aims to produce a handover interface design that improves and optimizes a wide range of experimental outcomes related to human performance: safety, situation awareness, trust, workload, acceptance, and usability. The work addresses many outcomes, as reducing the issue to a single outcome may not provide suitable solutions due to potential trade-offs and optimization problems. The design will achieve this by aligning with preexisting concepts in human team shift-handover practices, communicative principles from human–machine teamwork and the theory of distributed situation awareness. The HMI design will be generated through a four-step method of scoping the field, ...

Table of contents

• Cover
• Half Title
• Series Page
• Title Page
• Copyright Page
• Contents
• Preface
• Acknowledgements
• Authors
• Chapter 1 Introduction
• Section I Scoping the Issues and Solutions that Other Domains Face with Task Continuity
• Section II Pilot Testing These Concepts in Automated Driving
• Section III Designing New Interfaces and Interactions for Automated Vehicle Communication
• Section IV Testing and Validating a Novel Prototype
• Appendix A Cue Cards for Vocal Procedure—Chapter 5
• Appendix B HUD Slides for Final Design Solution
• List of References
• Index