ISO/TR 21974-1:2018

TECHNICAL ISO/TR
REPORT 21974-1
First edition
2018-11
Naturalistic driving studies —
Vocabulary —
Part 1:
Safety critical events
Études naturalistiques de conduite — Vocabulaire —
Partie 1: Événements critiques de sécurité
Reference number
ISO/TR 21974-1:2018(E)
©
ISO 2018

---------------------- Page: 1 ----------------------
ISO/TR 21974-1:2018(E)

COPYRIGHT PROTECTED DOCUMENT
© ISO 2018
All rights reserved. Unless otherwise specified, or required in the context of its implementation, no part of this publication may
be reproduced or utilized otherwise in any form or by any means, electronic or mechanical, including photocopying, or posting
on the internet or an intranet, without prior written permission. Permission can be requested from either ISO at the address
below or ISO’s member body in the country of the requester.
ISO copyright office
CP 401 • Ch. de Blandonnet 8
CH-1214 Vernier, Geneva
Phone: +41 22 749 01 11
Fax: +41 22 749 09 47
Email: copyright@iso.org
Website: www.iso.org
Published in Switzerland
ii © ISO 2018 – All rights reserved

---------------------- Page: 2 ----------------------
ISO/TR 21974-1:2018(E)

Contents Page
Foreword .iv
Introduction .v
1 Scope . 1
2 Normative references . 1
3 Terms and definitions . 1
Annex A (informative) Conceptual Framework for Categorizing Safety-critical events (SCEs)
in Naturalistic Driving Data . 5
Annex B (informative) Principles of Conflict Classification .11
Annex C (informative) Additional Conflict Attributes.28
Bibliography .52
© ISO 2018 – All rights reserved iii

---------------------- Page: 3 ----------------------
ISO/TR 21974-1:2018(E)

Foreword
ISO (the International Organization for Standardization) is a worldwide federation of national standards
bodies (ISO member bodies). The work of preparing International Standards is normally carried out
through ISO technical committees. Each member body interested in a subject for which a technical
committee has been established has the right to be represented on that committee. International
organizations, governmental and non-governmental, in liaison with ISO, also take part in the work.
ISO collaborates closely with the International Electrotechnical Commission (IEC) on all matters of
electrotechnical standardization.
The procedures used to develop this document and those intended for its further maintenance are
described in the ISO/IEC Directives, Part 1. In particular the different approval criteria needed for the
different types of ISO documents should be noted. This document was drafted in accordance with the
editorial rules of the ISO/IEC Directives, Part 2 (see www .iso .org/directives).
Attention is drawn to the possibility that some of the elements of this document may be the subject of
patent rights. ISO shall not be held responsible for identifying any or all such patent rights. Details of
any patent rights identified during the development of the document will be in the Introduction and/or
on the ISO list of patent declarations received (see www .iso .org/patents).
Any trade name used in this document is information given for the convenience of users and does not
constitute an endorsement.
For an explanation on the voluntary nature of standards, the meaning of ISO specific terms and
expressions related to conformity assessment, as well as information about ISO's adherence to the
World Trade Organization (WTO) principles in the Technical Barriers to Trade (TBT) see the following
URL: www .iso .org/iso/foreword .html.
This document was prepared by Technical Committee ISO/TC 22, Road vehicles, Subcommittee SC 39,
Ergonomics.
A list of all parts in the ISO 21974 series can be found on the ISO website.
Any feedback or questions on this document should be directed to the user’s national standards body. A
complete listing of these bodies can be found at www .iso .org/members .html.
iv © ISO 2018 – All rights reserved

---------------------- Page: 4 ----------------------
ISO/TR 21974-1:2018(E)

Introduction
A better understanding of driver behaviour is critical to future improvements in transportation safety.
Naturalistic Driving Studies (NDSs), which observe driver behaviour in a “natural” or uncontrolled
driving environment, offer unique insight into drivers’ typical behaviour under both “normal” driving
conditions and during the critical seconds before a crash or other safety-critical event (SCE). Much
of the value and power of NDSs lies in the video that is recorded of the driver and the environment
surrounding the vehicle. Although rich in information, video should be manually reviewed and coded
by trained data coders before it is scientifically analyzed. Given the potential for human error and
interpretation, coding protocols that are well-designed, thoroughly tested, and standardized across
studies are essential.
Naturalistic driving data sets, such as the Second Strategic Highway Research Program (SHRP 2), are
becoming more available to a diverse group of researchers. As a result, it is important to have a common
terminology for monitoring, coding, and analyzing data to allow research protocols to be replicated and
results to be compared across studies. In fact, the research community has called for the development of
fully tested, common coding protocols for use in NDSs. This document addresses that need by providing
a standard vocabulary for coding SCE characteristics in NDSs. The foundation for this document came
from the SHRP 2 naturalistic driving study annotation effort and subsequent revisions to the SHRP 2
[1]
dictionary to accommodate heavy vehicles (trucks and buses) . Substantial improvements have been
made to this document in both content and structure. However, large parts of this document are largely
verbatim with the original foundational documents.
It is recommended that vehicles in these studies be instrumented with at least a forward-looking view
and an in-vehicle view capturing the driver’s face and upper body. Rear- and side-facing views are
often helpful when interpreting conflicts that occur behind or next to the instrumented vehicle. A view
capturing the steering wheel, driver’s hands and/or dashboard can be helpful for additional analyses
(which are outside the scope of this document). However, equipment and labor costs may make these
additional views unfeasible. In addition, basic measurements of the kinematics of the instrumented
vehicle should be available, including at least longitudinal acceleration, lateral acceleration, and vehicle
speed. Other kinematic measurements that help assess conflict situations (if feasible) include brake and
throttle pedal application and/or pressure, proximity to and speed of surrounding non-instrumented
vehicles (e.g., via radar), latitude and longitude, and activation of key vehicle safety systems (e.g., anti-
lock brakes).
The main objective of this document is to define different types of SCEs based on a taxonomy of general
conflict classes and a set of basic variables for characterizing the events. The definitions supplied here
apply to data collected from light and heavy vehicles [i.e., category M and N according to Classification
and definition of power-driven vehicles and trailers: Council Directive 70/156/EEC (as amended by 92/53/
EEC), Annex 2]. However, this does not preclude the definitions specified in this document from, with
caution, being adapted for use with data collected from other vehicle types such as powered two-
wheelers or an infrastructure-based (rather than vehicle-based) data collection system.
This document does not address methods for identifying candidate SCEs (also referred to as triggers),
a complete set of annotations for these SCEs (e.g., driver distraction), or the recommended approach to
analyzing this data. This document also does not address the definition, extraction, or annotation of
controls. These controls, often called baselines in transportation literature, are selected for comparison
to events of interest in epidemiological as well as other analyses. Since the definition, extraction, and
annotation of these controls are highly dependent on the research question of interest, they have been
excluded from the scope of this document. The scope of the present document is graphically illustrated
in Figure 1.
© ISO 2018 – All rights reserved v

---------------------- Page: 5 ----------------------
ISO/TR 21974-1:2018(E)

Figure 1 — Document scope
vi © ISO 2018 – All rights reserved

---------------------- Page: 6 ----------------------
TECHNICAL REPORT ISO/TR 21974-1:2018(E)
Naturalistic driving studies — Vocabulary —
Part 1:
Safety critical events
1 Scope
This document defines terms and definitions commonly used for the annotation of video from NDSs
collected during real-world driving in an uncontrolled setting.
2 Normative references
There are no normative references in this document.
3 Terms and definitions
ISO and IEC maintain terminological databases for use in standardization at the following URL
addresses:
— ISO Online browsing platform: available at https: //www .iso .org/obp
— IEC Electropedia: available at http: //www .electropedia .org/
3.1
combined avoidance capacity
total response capacity of all responding conflict partners (3.3)
Note 1 to entry: The capacity of a responding conflict partner takes into account both the abilities and limitations
of the vehicle or road user as well as any environmental or infrastructural constraints, if present.
Note 2 to entry: Conflict partners that do not exhibit an observable response are not included in this construct.
3.2
conflict
situation where the trajectory(ies) of one or more road users or objects (conflict partner; 3.3) led to one
of three results: 1) a crash (3.4) or road departure (3.12), 2) a situation where an evasive manoeuvre(s)
(3.5) was required to avoid a crash or road departure, or 3) an unsafe proximity between the conflict
partners
Note 1 to entry: The key concept underlying the present framework is that of conflict.
Note 2 to entry: Three general classes of traffic conflict are of interest in naturalistic driving analyses: trajectory
conflict (3.2.1), single-vehicle conflict (3.2.2), and proximity conflict (3.2.3).
3.2.1
trajectory conflict
crash course between at least two conflict partners (3.3)
3.2.2
single-vehicle conflict
conflict (3.2) involving loss of vehicle control (e.g., horizontal and/or lateral skidding or rotation) or
proximity to the road edges (e.g., road departure; 3.12) rather than proximity to another entity
© ISO 2018 – All rights reserved 1

---------------------- Page: 7 ----------------------
ISO/TR 21974-1:2018(E)

3.2.3
proximity conflict
conflict (3.2) involving two or more entities that are not on a crash course but nevertheless come in
close temporal and/or spatial proximity to a crash (3.4)
3.3
conflict partner
any entity that is part of a conflict (3.2)
Note 1 to entry: This may include other vehicles (3.18), pedestrians (3.10), pedal cyclists (3.9), other non-motorists
(3.8), other road users, animals, and objects (including roadside barriers that exceed the ground clearance of the
affected vehicle).
Note 2 to entry: If a conflict partner (3.3) is present, then the conflict is either a trajectory conflict (3.2.1) or a
proximity conflict (3.2.3).
Note 3 to entry: Low roadside barriers (e.g., curbs) within the ground clearance of the vehicle are not considered
conflict partners.
3.4
crash
situation in which the subject vehicle (i.e., instrumented vehicle; 3.14) has any contact with at least one
other conflict partner (3.3) either on or off the trafficway (3.15), either moving or stationary (fixed or
non-fixed), that is observable or in which kinetic energy is measurably transferred or dissipated
Note 1 to entry: This excludes roadway (3.11) features meant to be driven over such as speed bumps and low
roadside barriers (curbs, medians, etc.) within the ground clearance limitations of the vehicle.
Note 2 to entry: A crash may also be a single-vehicle conflict (3.2.2) that includes at least one of the following
conditions: vehicle rollover, airbag deployment, injury, more than 90° degrees of horizontal vehicle rotation, or
all four tires leaving the trafficway.
3.5
evasive manoeuvre
any action performed by any conflict partner (3.3) to change its trajectory or speed in an attempt to
avoid or reduce the severity of a potential crash (3.4), avoid or reduce the severity of a road departure
(3.12), or regain vehicular control after a loss of control
Note 1 to entry: Examples include steering, braking, accelerating, running, stopping, or a combination of these.
Note 2 to entry: At least one of the manoeuvres exceeds normal vehicle control inputs.
3.6
metadata
information that provides a description about the structural content of its referenced dataset or the
methods with which the data were collected
Note 1 to entry: Examples of metadata include location (e.g., country, region) of data collection, sampling
methods, units of measure, etc.
3.7
naturalistic driving study
NDS
any driving study where research subjects are recruited to drive on public roads (not in a simulator or
on a test track), where there is no in-vehicle experimenter or confederate vehicles, and where driving
conditions are not experimentally controlled or manipulated
Note 1 to entry: Subjects are not instructed to drive differently than they normally would, and the data collection
instrumentation is unobtrusive.
Note 2 to entry: Typically, these studies last a minimum of several weeks for each subject and can go much longer.
2 © ISO 2018 – All rights reserved

---------------------- Page: 8 ----------------------
ISO/TR 21974-1:2018(E)

3.8
non-motorist
non-motorized conveyance
person is being transported by a non-motorized conveyance, other than a pedaled cycle (3.9)/a human-
powered device by which a non-motorist may move or may move another non-motorist
EXAMPLE Baby carriages, coasters, wagons, ice skates, roller skates, push carts, push scooters, skateboards,
skis, sleds, non-motorized wheel chairs, rickshaws, etc.
3.9
pedal cyclist
person on any type of manually propelled pedaled cycle, as either the driver or the passenger, including
bicycles, tricycles, and unicycles
Note 1 to entry: This category includes pedal cyclists holding on to a motorized vehicle.
Note 2 to entry: This category also includes power-assisted pedaled cycles that have limitations on speed (i.e.,
pedelecs). Pedaled cycles that are capable of propelling themselves and do not require pedaling for propulsion
are not included in this category (see “Motorcycle or moped”).
3.10
pedestrian
any person who is on or near a roadway (3.11) or a sidewalk, path, or other space that is contiguous
with a roadway, or on areas beside the roadway into which a vehicle can travel, and who is not in or on
either a motorized or a non-motorized conveyance
Note 1 to entry: This includes persons who are in contact with the ground, roadway, etc., but who are holding on
to a vehicle.
3.11
roadway
portion of a trafficway (3.15) that is designed and ordinarily used for vehicular travel, including all
designated or implied travel lanes (through lanes, turn lanes, acceleration and deceleration lanes), but
not shoulders, painted (whether usable or not), medians of any type, roadsides, gore areas, etc., that are
of a similar road surface to the parking lanes, parking areas, or driveways
3.12
road departure
conflict (3.2) in which the subject vehicle (3.14) is on a path toward and crosses, or is at risk of crossing,
the road edge
Note 1 to entry: The road edge is the outside edge of the shoulder (if present) or a physical raised curb or median
on the left or right side of the trafficway, including low barriers such as curbs and curb-style medians that are
within the ground clearance of the vehicle.
3.13
safety-critical event
SCE
conflict (3.2) or series of related conflicts that involves the subject vehicle (3.14) either alone or in
combination with another vehicle (3.18), pedal cyclist (3.9), pedestrian (3.10), object, or road edge
Note 1 to entry: This document describes the range of conflict types that may comprise an SCE, and an SCE may
be composed of a single conflict type or multiple simultaneous or sequential conflict types. Conflicts should be
non-intentional and non-premeditated (unplanned) by at least one conflict partner (3.3).
3.14
subject vehicle
vehicle (3.18) that has been instrumented and collects data
Note 1 to entry: Available videos will capture images from the subject vehicle’s perspective.
© ISO 2018 – All rights reserved 3

---------------------- Page: 9 ----------------------
ISO/TR 21974-1:2018(E)

Note 2 to entry: Kinematic data are typically available only from the subject vehicle or (in the case of radar data)
the subject vehicle’s perspective.
3.15
trafficway
any right-of-way designated for moving persons or property from one place to another, including the
surface on which vehicles normally travel (i.e., the roadway; 3.11), plus the shoulders, painted medians,
and painted gore areas at grade with the roadway
Note 1 to entry: The trafficway also includes parking lanes and parking areas (e.g., parking lots, driveways).
Note 2 to entry: The trafficway is bound by the outer edges of the shoulder or by raised roadside barriers (e.g.,
curb, guardrail, pylon) and thus does not include raised medians, grassy medians, sidewalks, etc.
3.16
trigger
data analysis method that includes the application of algorithms (e.g., acceleration thresholds) for
searching time series driving data for key situational characteristics or thresholds in order to identify
potential events of interest (e.g., kinematic threshold for potential SCEs; 3.13)
3.17
urgent response
urgency of response
situation in which the required evasive manoeuvre (3.5) approaches the combined avoidance capacity
(3.1) of the responding conflict partners (3.3) required to prevent a crash
Note 1 to entry: Both the time available for response and the required response magnitude (e.g., level of
deceleration) should be considered.
Note 2 to entry: While it may be possible to quantify the required urgency of the evasive manoeuvre (see Annex A
for more discussion), the methods for doing so are not within the scope of this document. Quantification of the
urgency is not required in order to use the concept as described in this document.
Note 3 to entry: The urgency of the response is independent of any evasive manoeuvres that are actually
performed.
3.18
vehicle
any motorized means of transportation, excluding pedelecs, which are included in the pedal cyclist
definition (3.9)
4 © ISO 2018 – All rights reserved

---------------------- Page: 10 ----------------------
ISO/TR 21974-1:2018(E)

Annex A
(informative)

Conceptual Framework for Categorizing Safety-critical events
(SCEs) in Naturalistic Driving Data
A.1 Introduction
Safety-critical events (SCEs) of interest in naturalistic driving studies (NDSs) traditionally include
crashes and near-crashes, but also less-severe but still safety-relevant non-crash events, such as crash-
relevant conflicts and proximity conflicts. The main objective of this document is to define and rank
different types of SCEs based on a taxonomy of general conflict classes and a set of basic variables
for characterizing the events. The objective of this Annex is to outline the conceptual basis for these
definitions.
The key concept underlying the present framework is that of traffic conflict. There are at least three
general classes of traffic conflict that are of interest in naturalistic driving analyses. First, a trajectory
[1]
conflict, which is the main conflict class traditionally studied (e.g., Reference ), is defined in terms of a
crash course between at least two conflict partners. Second, a single-vehicle conflict is defined in terms
of loss of control or proximity to the road edges rather than by proximity to another conflict partner.
Third, a proximity conflict involves two or more conflict partners that are not on a crash course but
nevertheless come in close temporal and/or spatial proximity to a crash. Importantly, these general
conflict types are not mutually exclusive and may coexist in a single SCE. Thus, an SCE may include
several conflicts occurring in a series.
Furthermore, these conflict classes are characterized by two types of criteria. First, the outcome criteria
state that conflicts result in some level of criticality or severity, and that conflicts within each class
may be graded based on these severity levels. However, the definitions of these severity levels differ
between the conflict classes, as further described below. Second, the intentionality criterion states that
conflicts are always non-intended by at least one of the conflict partners. The following section further
describes these two criteria, and the remainder of the Annex describes how they apply to each of the
three conflict classes.
A.2 General conflict criteria
A.2.1 Outcome criteria
The basic criterion for an event to count as a crash is that it involves contact with another entity that
extends above the ground clearance of the subject vehicle that is observable and in which kinetic energy
is measurably transferred or dissipated. In the case of a single vehicle crash, the vehicle departs the
road with all four tires, rolls over, results in an air bag deployment, or involves occupant injury. The
crash can then be further characterized in terms of outcome criteria such as delta-V (the change in
velocity during impact) and/or other observable outcomes such as injuries or property damage. In the
context of naturalistic driving analysis, the choice of crash-severity criteria to be applied to a given
conflict depends mainly on practical rather than conceptual considerations, in particular the types of
outcomes that can be reliably observed in naturalistic data.
For non-crashes, the specific outcome criteria differ between the main conflict classes, but the common
characteristic is that there is some degree of kinematic proximity to crash impact, or, in the case of
single-vehicle conflicts, some degree of loss of control or proximity to the road edges. The specific
criteria for each conflict class are further discussed below.
© ISO 2018 – All rights reserved 5

---------------------- Page: 11 ----------------------
ISO/TR 21974-1:2018(E)

A.2.2 Intentionality criterion
The intentionality criterion requires that the conflict (i.e., crash impact or kinematic proximity) was
not intended by at least one of the conflict partners. This criterion is important to exclude situations
where a driver (or, in the case of trajectory and proximity conflicts, all conflict partners) intentionally
produces a situation that satisfies the outcome criteria. For example, this may involve a driver leaving
the road on purpose to take a shortcut. Importantly, it should be noted that the action leading up to
a true conflict (e.g., initiating a pass on a blind corner) may have been intentional, but as long as the
outcome [e.g., a small time-to-collision (TTC) to an oncoming vehicle] was not intended by at least one
of the conflict partners, the event should be considered a conflict.
A.3 Trajectory conflict
In traditional conflict theory, a conflict is defined as “a situation where two or more conflict partners
approach each other in time and space to such extent that a crash is imminent if their movements
[2]
remain unchanged ”. Thus, the defining features of traditional conflicts are (1) that two or more
conflict partners (which may be vehicles, other road users, or stationary objects) enter into a collision
course with each other where (2) collision is imminent. This general situation is referred to as a trajectory
conflict, which includes the main types of SCEs traditionally analyzed in NDSs, such as crashes, near-
crashes, and crash-relevant conflicts.
Given this conceptualization of a trajectory conflict, a key further issue is to characterize the degree
of imminence of a crash. Here, imminence is conceptualized in terms of the urgency of the conflict
situation. Urgency is here defined as given the point of time of the avoidance manoeuvre, what is the
required magnitude of the avoidance response relative to the possible (or available) avoidance response
magnitude of the involved responding conflict partners. More specifically, a response is urgent when
the magnitude of the evasive manoeuvre required to avoid a crash approaches the combined avoidance
capacity of all responding conflict partners in the given situation.
There are two key concepts here: the required evasive manoeuvre magnitude, and the combined
avoidance capacity of all responding conflict partners. The required evasive manoeuvre magnitude is
discussed in greater detail in the paragraphs that follow, but in short it represents the “minimum” that
should be done to avoid a crash. It is dependent on the relative speed of the conflict partners and the
angle of their approach. It is independent of the actual response(s) performed and may be of greater (in
the case of crashes) or lesser (in the case of non-crash conflicts) magnitude than the actual response(s).
Avoidance capacity is here defined as the ability of a conflict partner to execute an evasive manoeuvre,
where an evasive manoeuvre is any action performed to avoid a potential crash by changing the
trajectory or speed, such as steering, braking, accelerating, running (pedestrian and animals), or
stopping. The combined avoidance capacity refers to the combined ability of all responding conflict
partners to change their trajectories to avoid a crash. It is also independent of the actual response(s)
performed, but dependent on the characteristics both of the responding conflict partners themselves
(e.g., brake condition, tire tread, vehicle specifications, braking and steering capabilities, pedestrian
mobility, etc.), and the environment in which the conflict takes place (e.g., road surface condition,
presence of other road users or obstacles). For example, the combined avoidance capacity of a vehicle
that brakes and/or swerves as it heads toward a pole is less than the combined avoidance capacity
in a situation where the same vehicle approaches a pedestrian and both conflict partners respond to
the situation, everything else being equal (since the pedestrian but not the pole can move). Finally, a
responding conflict partner is defined as a conflict partner exhibiting a visible response to the conflict.
If a conflict partner does not exhibit such a response, it is excluded from the conceptualization of
“combined avoidance capacity” for the given conflict. Thus, if the pedestrian in the example above does
not move, then the combined avoidance capacity is the same as in the situation with the pole.
A useful way to think about the urgency of a situation is in terms of the proximity to the point-of-
no-return (PONR). The PONR refers to the point in time during the conflict envelope when the crash
becomes unavoidable given the maximum avoidance capacities of the responding conflict partners.
Thus, for example, for a given set of kinematics (locations, speeds, and accelerations) of the involved
conflict partners, the urgency of the situation increases with reduced road friction (thus reducing
6 © ISO 2018 – All rights reserved

---------------------- Page: 12 ----------------------
ISO/TR 21974-1:2018(E)

the avoidance capacities of the conflict partners). Consequently, the PONR will be reached earlier,
and the situation will be more urgent, when the road is wet or icy compared to when it is dry. This
conceptualization of urgency may be used as the basis for characterizing different types of non-crash
conflicts. For example, near-crashes (NCs) may be defined as situations with high urgency. That is,
the magnitude of the required evasive manoeuvre approaches the maximum combined avoidance
capability of the responding conflict partners (see B.2.1.4 and B.2.1.5). In other words, a near crash
corresponds to a situation with a small margin to the PONR. By contrast, a crash-re
...