Ontology Design for Task Allocation and Management in Urban Search

and Rescue Missions

Elie Saad

, Koen V. Hindriks

and Mark A. Neerincx

1,2

Department of Intelligent Systems - Interactive Intelligence Group, Delft University of Technology, Delft, The Netherlands

The Netherlands Organization for Applied Scientiﬁc Research (TNO), Soesterberg, The Netherlands

Keywords:

Task Management, Human Robot Collaboration, Ontology, Urban Search and Rescue.

Abstract:

Task allocation and management is crucial for human-robot collaboration in Urban Search And Rescue

response efforts. The job of a mission team leader in managing tasks becomes complicated when adding

multiple and different types of robots to the team. Therefore, to effectively accomplish mission objectives,

shared situation awareness and task management support are essential. In this paper, we design and evaluate

an ontology which provides a common vocabulary between team members, both humans and robots. The

ontology is used for facilitating data sharing and mission execution, and providing the required automated

task management support. Relevant domain entities, tasks, and their relationships are modeled in an ontology

based on vocabulary commonly used by ﬁremen, and a user interface is designed to provide task tracking

and monitoring. The ontology design and interface are deployed in a search and rescue system and its use is

evaluated by ﬁremen in a task allocation and management scenario. Results provide support that the proposed

ontology (1) facilitates information sharing during missions; (2) assists the team leader in task allocation and

management; and (3) provides automated support for managing an Urban Search and Rescue mission.

1 INTRODUCTION

After a disaster, such as a hurricane or an industrial

accident, ﬁreﬁghters arrive on site with different

types of robots to perform Urban Search And Rescue

(USAR) response efforts (Murphy, 2004). During

these efforts, human-robot team leaders have to act

fast and allocate tasks to ﬁremen (robot operators,

inﬁeld rescuers, etc.) to assess the situation and save

potential victims. Firemen will then collaborate with

robots, such as unmanned ground vehicles (UGVs)

and aerial vehicles (UAVs), to execute these tasks.

Such human-robot collaboration sets new challenges

concerning task allocation and management (Murphy

et al., 2008; Lewis et al., 2010).

In this race against time, any wrong decision when

allocating or executing a task may cause additional

damages and risk the lives of both victims and

rescuers. For example, after an earthquake hit Mexico

City in 1985 when limited resources for inspecting a

disaster site were available, many rescuers died while

executing USAR tasks. Of these rescuers, 65 drowned

in the area where they were assigned to search for

victims (Casper and Murphy, 2003).

Nowadays, the amount of resources for

exploration and reconnaissance has increased, in

particular because of the availability of rescue robots

(Murphy, 2014) with various types of sensors and

detectors. The downside of adding robots to the mix,

however, is that this also increases the workload of

the team leader who needs to select which resource(s)

to use for performing a given task. Moreover, the

use of robots leads to a substantial increase of the

heterogeneous data gathered from the disaster site

(point clouds generated by cameras and lasers, etc.)

which needs to be analyzed and taken into account

when deciding on the appropriate actions to take. For

example, if a robot detects a gas leak close to a ﬁre,

the raw data should be analyzed by verifying the gas

density and its proximity to ﬁre, before deciding to

send ﬁreﬁghters to that area.

An USAR team leader needs to be aware of

the current situation (Riley and Endsley, 2004) and

take many elements into consideration in order to

allocate tasks effectively to a human-robot team. Key

elements, for example, include (1) the available actors

(operators, rescuers, UGVs, UAVs) and their current

state (location, battery level, workload, etc.); (2) the

capabilities (sensory, locomotion, communication)

and devices (thermo and waterproof cameras, ﬁre and

622

Saad, E., Hindriks, K. and Neerincx, M.

Ontology Design for Task Allocation and Management in Urban Search and Rescue Missions.

DOI: 10.5220/0006661106220629

In Proceedings of the 10th International Conference on Agents and Artiﬁcial Intelligence (ICAART 2018) - Volume 2, pages 622-629

ISBN: 978-989-758-275-2

gas detectors, etc.) that actors have or are equipped

with; (3) a map of the disaster area to keep track of

what has been explored so far, including the detected

points of interest (POIs), such as hazardous materials

and victims; and (4) the communication options

available, including their range so as not to lose

contact with the team.

To address the aforementioned challenges, a

sophisticated human-robot task management model

is needed to support (1) building and maintaining

shared situation awareness (Endsley, 1995); (2)

preventing overload of operators and inﬁeld rescuers

(Neerincx, 2003); and (3) handling interoperability

(Missikoff and Taglino, 2004) and interdependencies

(Johnson et al., 2014) within a human-robot team.

We focus on creating a knowledge representation

for mapping data to a uniﬁed semantics (Sheth,

1999; Xu and Zlatanova, 2007) which is shared

between humans and robots. We propose the use

of an extensive ontology (Schlenoff and Messina,

2005; Mescherin et al., 2013) for conceptually

representing USAR domain entities and tasks, and

their relationships. This conceptualization is useful

for task allocation and management as it (1)

captures relevant domain aspects; (2) facilitates

communication and information sharing; (3) provides

reasoning support throughout a mission; and (4)

improves human-robot collaboration.

This paper describes the ontology we designed

for assisting ﬁremen and the team leader in mission

control. It provides a common vocabulary to be used

by both, ﬁremen and robots, and automated support

for task management in particular for the team leader.

Throughout the paper we use a scenario from the

TRADR project, short for ’Long-Term Human-Robot

Teaming for Robot Assisted Disaster Response’, see

(Kruijff-Korbayov

a et al., 2015), to illustrate the main

concepts. The scenario involves a reconnaissance

sortie and barrel inspection which requires a team

composed of a ﬁremen who is the team leader, robot

operators and inﬁeld rescuers, and robots including

UGVs and UAVs. Team members receive tasks from

the team leader to scout the disaster area and gather

more information about a barrel that has been spotted.

The main contributions of this paper are (1) the

design of an ontology; (2) the design of a user

interface which displays (parts of) the ontology and

is part of automated task support that assists the team

leader during a mission; and (3) the evaluation of the

ontology and automated task support by ﬁremen in a

task allocation and management scenario.

This paper is organized as follows. In Section 2

we review related work. In Section 3 we present the

development and structure of the task management

ontology. In Section 4 we present the user interface

which displays parts of the ontology and is part of

a larger search and rescue system. In Section 5 we

discuss the evaluation of the designed ontology based

on its use during a reconnaissance sortie use case and

interviews with ﬁreﬁghters. In Section 6 we conclude

the paper and give directions for future research.

2 RELATED WORK

Ontologies are widely used to represent domain

knowledge and facilitate information sharing in

many applications (Rivero et al., 2013; Missikoff

and Taglino, 2004). According to (Liu et al.,

2013a), existing crisis oriented ontologies describe

concepts and characteristics related to a single subject

area (type of disasters, geography, meteorology,

etc.). Such ontologies are designed to address the

requirements of a speciﬁc application.

In USAR robotics ontologies, work has been

done on the cooperation between autonomous or

semi-autonomous multi-robot teams (Liu et al.,

2013b). For example, the robot ontology suggested

by Schlenoff and Messina, 2005, is centered on

representing the concepts about robots and their

capabilities when assisting in USAR missions. This

ontology is very relevant, but differs from our focus

on providing task management support for a team

leader.

Other robotics ontologies focus on the robot

and its interaction with a speciﬁc environment.

For example, (Jacobsson et al., 2016) proposes an

ontology for industrial use and (Li et al., 2017) for

underwater robots. The KnowRob ontology (Tenorth

et al., 2013) offers a set of ontologies to model robots

and their capabilities and actions. The OpenRobot

(Lemaignan et al., 2010) ontology (ORO), which

shares many concepts with the KnowRob ontology,

is focused on robot interaction with humans, but

assumes that robots are completely autonomous.

Given the current state of the art in USAR robotics,

we focus instead on remotely operated robots.

3 ONTOLOGY DESIGN

In this section, we ﬁrst discuss the development

process we followed to build the ontology introduced

in this paper. Then we present the overall structure of

the ontology and discuss the main concepts that have

been included to support task management.

Ontology Design for Task Allocation and Management in Urban Search and Rescue Missions

623

3.1 Ontology Development Process

The development process we followed to build our

ontology is adapted from (Simperl and Tempich,

2006). This iterative process consists of different

phases, as illustrated in Figure 1. First, we analyzed

the task management system requirements in search

and rescue domain by interviewing ﬁreﬁghters

and reviewing the literature and state-of-the-art

ontologies. Second, the required domain entities

and their relationships are conceptualized based on

common vocabulary used by ﬁremen. Third, the

ontology is implemented as RDF triples in the OWL

2 Web Ontology Language syntax

and visualized

through the system’s user interfaces. Fourth, the

modeled ontology is evaluated by ﬁremen using

search and rescue use case scenarios. Lastly, in

the maintenance phase, the ontology is reﬁned and

extended with new concepts and entities needed for

addressing the requirements of additional use cases.

Figure 1: Ontology development process adapted from

(Simperl and Tempich, 2006).

3.2 Overall Ontology Structure

The ontology introduced in this paper has been

designed to be part of a larger ontology (Bagosi

et al., 2016) that is used in the TRADR system,

a European project for search and rescue response

efforts

. We aimed to make the design of the ontology

(1) ﬂexible and extensible, to be able to easily append

new components, for e.g. covering more use cases;

(2) reusable, so it can be applied in different types of

missions; and, perhaps most importantly, (3) readily

understandable by ﬁremen, the key rescuers in our

domain, in order to facilitate task allocation and

management.

To this end, we summarized and grouped the

knowledge gathered into multiple modules in order

https://www.w3.org/TR/owl2-syntax/

http://www.tradr-project.eu

to append additional components as we extend the

core ontology. Four of these modules are relevant and

part of the task management ontology, as illustrated

in Figure 2. The ActorModule groups the agents,

both humans and robots, as actors along with their

properties such as roles and team afﬁliations. The

CommunicationModule groups all concepts needed to

facilitate data gathering and exchange between team

members, such as communication events (messages,

notiﬁcations, etc.), media types (video, photo,

audio, etc.) and data gathering devices (infra red

sensor, camera, etc.). The EnvironmentModule

groups the environmental events (hurricane, ﬂood,

chemical leakage, etc.) and structures. Finally, the

MissionModule contains the concepts and entities

needed when setting up and planning a mission,

including a taxonomy of tasks and POIs.

3.3 Modeling and Requirements

The detailed design of our ontology has been

based on our discussions and interviews with

ﬁreﬁghters, experts in the ﬁeld, and also has

been inspired by Robin R. Murphy’s research on

rescue robotics (Murphy, 2014). Our ontology is

aimed at providing a common vocabulary which

is useful for task management by facilitating data

sharing and communication between team members.

The concepts represented in our ontology therefore

include all the relevant entities and information

categories that are needed for task allocation and

management during USAR sorties. The latter are

executed using remotely operated robots. We brieﬂy

discuss the key concepts that have been included

in the different modules and their relationships, as

illustrated in Figure 2.

3.3.1 Actor Module

The actor ontology represents the human and robot

actors along with their properties. The actors are

resources used for responding to the disaster. By

representing the roles, status, capabilities, and related

concepts in the ontology, the latter can provide a basis

for automated support for task management. The

system, for example, can reason which actors might

be suitable for performing a speciﬁc task.

• Actor Roles and Teams - During an USAR

mission, human and robot actors collaborate for

executing the tasks assigned by the team leader.

To know who will do what, human actors have

different roles (UGV operator, inﬁeld rescuer,

etc.) as do robot actors (e.g., ground or aerial

explorers). We have based the model of the

ICAART 2018 - 10th International Conference on Agents and Artiﬁcial Intelligence

624

Figure 2: Class diagram illustrating the different ontology modules (Actor, Communication, Environment and Mission) and

some of their entities and interdependencies.

role concept on a TRADR human-robot team

which we believe is sufﬁciently general to apply

more generally. It has moreover been validated

by ﬁremen from three different countries that

participate in the project (Italy, Germany, and The

Netherlands). Teams are composed of a team

leader, UGV and UAV tele-operated robots and

their corresponding operators, inﬁeld rescuers,

robot mechanics and safety ofﬁcers.

• Actor properties - In addition to roles, each actor

has a status (idle, on the move, etc.). Human

actors have a workload, which is an indicator of

an actor’s task load during a mission. And robot

actors have battery readings, which indicate the

battery percentage at each moment in time. Such

properties help the team leader when assigning

tasks by showing which actor is available for

performing a given task.

• Actor Capabilities - Actors have capabilities

related to their skills (paramedic, etc.) and the

devices they are equipped with (infrared camera,

gas detector, etc.). Knowing the capabilities of

actors helps in providing automated support and

suggesting available actors to the team leader

when assigning a task.

3.3.2 Communication Module

In a USAR mission, gathering and sharing relevant

information is important for improving situation

awareness and facilitating task management. This

requires a communication network between actors.

• Communication devices - In this category we

model the devices needed for gathering and

communicating data between members, such as

electronic and sensing devices (infra red sensors,

thermo cameras, network devices, etc.).

• Communication events and media types -

Relevant information is shared using different

types of communication events (notiﬁcations,

messages, etc.) and media types (audio, photo,

text, etc.). This is helpful for keeping the team

leader and other members aware of the state

of a mission and alert them when something

unexpected occurs while executing a task.

3.3.3 Environment Module

Environmental objects and events in a disaster area

need to be inspected or handled by performing

different tasks, and therefore are important to

represent in a task management ontology.

• Environmental events - these include the events

which have happened or can occur while scouting

a disaster site and which need to be monitored and

handled by rescuers such as explosions, ﬁres, etc.

• Environmental objects - these include concepts

for representing structures, barrels, etc., and that

can be present in the disaster area.

3.3.4 Mission Module

Allocating and managing tasks helps in planning and

monitoring the progress of a mission. This requires an

overall view of the situation which includes, among

others, the location of active actors and of what was

Ontology Design for Task Allocation and Management in Urban Search and Rescue Missions

625

discovered so far along with the tasks assigned and

their progress.

• Tasks and relevant properties - Throughout a

mission, the team leader assigns tasks to available

actors by specifying the task objective or POI, and

providing a clear description. To monitor tasks

and track their progress, additional properties

have been included such as status (in progress,

completed, etc.) and priority. Moreover, each task

has a list of required capabilities such as sensing,

locomotion and communication, which deﬁnes to

which actor(s) the task can be allocated.

• Points of interest (POIs) - This category includes

the entities which can be detected while exploring

a disaster site and are meaningful for improving

situation awareness when assigning tasks. Each

POI has a type (victim, ﬁre, gas, hazards, etc.)

and a location. These properties help in knowing

which actors to send, where they should go and

what might be the risks involved.

4 USER INTERFACE

In order to be able to use and deploy the task

management ontology discussed in Section 3, two

different GUIs have been designed. These GUIs

integrate and provide support for task management in

the search and rescue tactical display system (TDS).

The TDS is used for tracking the disaster area and has

been developed to assist USAR teams in the TRADR

project. It contains a map of the disaster site showing

the location of actors and the detected POIs (victims,

Figure 3: Task Editor for creating and editing mission tasks.

ﬁres, chemical objects, etc.). It also provides

support for assessing the disaster site situation and for

gathering relevant information about it.

To allocate new tasks or edit existing ones, the

team leader uses a task editor, as shown in Figure 3. In

this editor, the team leader deﬁnes the task properties

including (1) a task type (search, go to, take photo,

etc.); (2) a POI which deﬁnes the task’s objective;

(3) a priority; (4) a status (pending, in progress,

completed, etc.); (5) a description containing speciﬁc

details or guidelines for the operators; (6) a list of

required capabilities which are automatically selected

by the system depending on the task type and can be

modiﬁed by the user; (7) a required battery level; (8)

a required workload; and (9) an actor from the list of

available actors suggested by the system.

Figure 4: Task Manager for tracking and monitoring tasks.

A second GUI provides the team leader with a task

manager interface (Figure 4) which has been designed

to enable the team leader to track and monitor the

progress of assigned tasks. For each task, the GUI

displays its description, assigned actor, priority and

status, to provide the team leader with an overview

of the execution progress. Mission actors can track

the progress of their tasks in the main display system

which shows the task name, objective and status.

The latter property is continuously updated by actors

throughout the execution process.

For every new mission, the main ontology is

initialized and loaded in a central repository (we

use Stardog triple stores

which provide support

for querying, inferencing and manipulating the

knowledge base stored in the repository based

on the semantics deﬁned by our ontology). To

ensure this repository maintains an up-to-date state

of a mission, we developed and use semantic

modelers to continuously update the database with

http://www.stardog.com

ICAART 2018 - 10th International Conference on Agents and Artiﬁcial Intelligence

626

new knowledge acquired during a mission. These

modelers map raw sensor data (e.g., point cloud, GPS

coordinates, etc.) onto ontological concepts (POIs,

locations, etc.) and store it in the repository.

The mapped data is then used to display and

update meaningful information for monitoring the

progress of a mission on TDS. It is also used to reason

about the represented world and generate notiﬁcations

related to the task being executed. The aim is to

manipulate the gathered knowledge for (1) improving

shared situation awareness; and (2) assisting the team

leader in its job of assigning tasks by providing

automated support. For example, when creating a task

for sending an UGV on site to take a photo of a POI,

the system queries the knowledge base to display the

list of available human actors who operate a UGV

equipped with a high-resolution camera and having

enough battery life. Whereas when creating a task

for picking up a sample to be analyzed, querying the

knowledge base returns the human actors operating a

UGV equipped with a robotic arm.

Figure 5: Activity diagram for assigning and executing

mission tasks.

Throughout a mission, the team leader will

continuously add new tasks or update existing ones

(description, priority, etc.). The activity diagram

in Figure 5 shows the workﬂow for assigning and

executing a task. First, the team leader assigns a task

to an actor. Then, the actor can accept it and start

the execution or can abort it when facing technical

issues (e.g., robot errors). When the task is executed,

the actor sets its status to awaiting acknowledgement

using the task manager. If the result is accepted,

the team leader sets the task status to completed.

Otherwise, it will be reassigned or canceled.

5 EVALUATION AND RESULTS

To evaluate the use of our ontology in USAR

missions, we evaluated it as part of the bigger

TRADR system while executing a use case scenario

that involves a reconnaissance sortie and the

inspection of a barrel. The scenario was executed by

ﬁremen teams at the ﬁre department training facility

located in Rozenburg, The Netherlands. After each

sortie, we interviewed the ﬁreﬁghters team and their

leader to obtain their feedback about the ontology (its

concepts) and its use for displaying task management

related information and content in the task editor and

management user interfaces.

5.1 Use Case Scenario

The scenario is based on an industrial accident where

an explosion has occurred on site. A team is sent

to (1) search for human victims; (2) gather more

information about the site; and (3) inspect the area

for the presence of chemical hazards and leakages.

First, the team leader has to assign a task to a UAV

operator to scout the disaster area. While executing

the task, the operator will spot an unidentiﬁed barrel

and should notify the team leader of this. When

the team leader is informed about the barrel, a task

should be assigned to an actor operating a UGV with

a high-resolution camera to inspect the barrel and

take a photo of it. After receiving the photo of the

barrel and analyzing it, the team leader should assign

a task to the actor operating a UGV with a robotic

arm to close the barrel opening and prevent potential

chemical leaking. All tasks will be inserted in the

Task Manager GUI, as shown in Figure 4, and actors

are made aware of the tasks assigned to them.

5.2 Execution

During two days, three ﬁremen teams alternated to

practice the use case scenario. They received a quick

introduction of the TRADR system and its interfaces

for about 20 minutes before starting the execution.

At the beginning of each mission, our ontology

was instantiated and loaded with the required entities.

In our use case, these entities include (1) four human

actors where one has a team leader role, two with

UGV operator role and one with UAV operator role;

(2) three robot actors where one is a UAV and

Ontology Design for Task Allocation and Management in Urban Search and Rescue Missions

627

two are UGVs; (3) environmental objects such as a

barrel and debris; (4) points of interest (POIs) such

as ﬁre and gas. Throughout a mission, the team

leader used the task management interfaces to allocate

tasks and monitor their progress. These interfaces

used ontological reasoning to provide the leader with

automated support by suggesting available actors for a

given task and generating notiﬁcations when needed.

After executing each sortie, an informal interview

took place with the ﬁremen in order to (1) verify that

they were satisﬁed with the support and interaction

offered by the system; and (2) estimate their level of

understanding of the ontological concepts represented

and shown in the user interfaces (i.e., did the ontology

provide a common vocabulary that ﬁremen could

understand?). Additional interviews took place with

the team leaders to check whether the designed

ontology ﬁts their needs when managing tasks and

executing the missions.

5.3 Results

The three ﬁremen that played the role of team leader

indicated that the Task Manager GUI is easy to use

and the ontology concepts are clear and easy to grasp.

They each said that the task assignment support was

intuitive to use and operated as expected. Even so

team leaders also indicated that they needed time to

get used to it (they had to switch from their usual

practice of taking notes on paper to using the user

interface).

Task actors did not bother to manually change

the task status when executing them. The reason is

that, in a real mission (as mentioned by ﬁremen), the

team leader assigns tasks and the operators perform

it. They only report back to the team leader and

change the task status when they ﬁnish the task or

whenever they encounter a problem during execution.

Therefore, it is suggested that the task status should

also be automated somehow by the system.

Furthermore, team leaders indicated that the

task editor should be simpliﬁed. They suggested

to provide default values to some of the ﬁelds,

especially those related to robots, and hide them when

creating a new task. These include the task priority,

status, required capabilities, required battery level and

required workload. Setting default values to these

ﬁelds allows the system to provide automated support

(1) by suggesting to the team leader the appropriate

actors who can execute a given task; and (2) by

generating appropriate notiﬁcations when a task is

wrongly assigned or cannot be executed, which helps

the team leader when monitoring task progress.

6 CONCLUSION

When planning and executing an USAR mission,

the team leader needs to efﬁciently allocate tasks to

team members for assessing the situation and rescuing

potential victims. The leader’s job is time-critical

and complex which requires, among other things, an

awareness of the current situation and the knowledge

of the team members capabilities and their actual

status. Using an ontology for assisting the team

leader in task allocation and management provides

a common vocabulary between team members, both

humans and robots. The ontology is useful for

(1) facilitating data sharing; (2) improving shared

situation awareness; and (3) providing automated

support in the task management process.

This paper introduced part of the ontology

developed for TRADR search and rescue project.

The ontology is focused on facilitating human-robot

collaboration by means of providing automated

support for task allocation and management

during USAR missions. It is evaluated using a

reconnaissance sortie and barrel inspection use case.

The evaluation shows that the ontology constitutes

a good basis for providing automated support to

assist a team leader in mission planning and task

management.

The main contributions have been the design of

the ontology and related user interfaces, as well as an

evaluation in a search and rescue project scenario.

6.1 Directions for Future Research

Our results helped us gain a better understanding of

the needs of ﬁremen in general and in particular of

the team leader in USAR missions. The following

points need to be taken into consideration and require

further analysis. It has become clear that ﬁreﬁghters

need more training to use advanced tools based

on ontologies and automated support. It remains

to be seen how we can further simplify system

support and whether this can be achieved by further

automation. It will be interesting to design, develop,

and evaluate additional automated support related to

task status updates. The aim should be to further

decrease the workload of ﬁreﬁghters and prevent

system automation to feel as a burden.

More evaluation, moreover, is needed and

additional use cases should be designed and used

for testing and evaluation purposes. Additional use

cases may reveal potential gaps not yet covered by

our ontology and provide new insights in what is

needed to automate task management support. We are

particularly interested in verifying whether our task

ICAART 2018 - 10th International Conference on Agents and Artiﬁcial Intelligence

628

taxonomy is sufﬁcient for speciﬁc subtasks in such

use cases. In any case, we expect more reﬁnements

will be needed for modeling robot capabilities (e.g.,

robot arm manipulation). Also, our ontology does not

yet provide support for robot-robot interaction, fully

autonomous robot operations, underwater operations,

and, for example, issues such as network resilience.

The aforementioned suggestions and extensions will

also require an exhaustive user evaluation to cover

more parts of the ontology in different scenarios.

ACKNOWLEDGEMENTS

This work was supported by European Union’s

Seventh Framework Programme for research,

technological development and demonstration under

the TRADR project No. FP7-ICT-609763.

REFERENCES

Bagosi, T., Hindriks, K. V., and Neerincx, M. A. (2016).

Ontological reasoning for human-robot teaming in

search and rescue missions. In 11th ACM/IEEE HRI,

pages 595–596.

Casper, J. and Murphy, R. R. (2003). Human-robot

interactions during the robot-assisted urban search

and rescue response at the world trade center. IEEE

Transactions on Systems, Man, and Cybernetics, Part

B (Cybernetics), 33(3):367–385.

Endsley, M. (1995). Toward a theory of situation awareness

in dynamic systems. Human Factor, 37:32–64.

Jacobsson, L., Malec, J., and Nilsson, K. (2016).

Modularization of skill ontologies for industrial

robots. In Proceedings of ISR 2016: 47st

International Symposium on Robotics, pages 1–6.

Johnson, M., Bradshaw, J. M., Feltovich, P. J., Jonker,

C. M., van Riemsdijk, M. B., and Sierhuis, M.

(2014). Coactive design: Designing support for

interdependence in joint activity. J. Hum.-Robot

Interact., 3(1):43–69.

Kruijff-Korbayov

a, I., Colas, F., Gianni, M., Pirri, F.,

de Greeff, J., Hindriks, K., Neerincx, M.,

Ogren,

P., Svoboda, T., and Worst, R. (2015). Tradr

project: Long-term human-robot teaming for robot

assisted disaster response. KI - K

unstliche Intelligenz,

29(2):193–201.

Lemaignan, S., Ros, R., M

osenlechner, L., Alami,

R., and Beetz, M. (2010). Oro, a knowledge

management platform for cognitive architectures in

robotics. In 2010 IEEE/RSJ International Conference

on Intelligent Robots and Systems, pages 3548–3553.

Lewis, M., Wang, H., Chien, S.-Y., Scerri, P.,

Velagapudi, P., Sycara, K., and Kane, B.

(2010). Teams organization and performance in

multi-human/multi-robot teams. In 2010 IEEE

International Conference on Systems, Man and

Cybernetics, pages 1617–1623.

Li, X., Bilbao, S., Martin-Wanton, T., Bastos, J., and

Rodriguez, J. (2017). Swarms ontology: A common

information model for the cooperation of underwater

robots. Sensors (Basel, Switzerland), 17(3):569–588.

Liu, S., Brewster, C., and Shaw, D. (2013a). Ontologies

for crisis management: a review of state of the art

in ontology design and usability. In Proceedings of

the 10th International ISCRAM Conference, pages

349–359, Baden-Baden, Germany.

Liu, Y., Nejat, G., and Vilela, J. (2013b). Learning

to cooperate together: A semi-autonomous control

architecture for multi-robot teams in urban search and

rescue. In IEEE International Symposium on Safety,

Security, and Rescue Robotics (SSRR), pages 1–6.

Mescherin, S. A., Kirillov, I., and Klimenko, S. (2013).

Ontology of emergency shared situation awareness

and crisis interoperability. In 2013 International

Conference on Cyberworlds, pages 159–162.

Missikoff, M. and Taglino, F. (2004). An Ontology-based

Platform for Semantic Interoperability, pages

617–633. Springer, Berlin, Heidelberg.

Murphy, R. R. (2004). Human-robot interaction in rescue

robotics. IEEE Transactions on Systems, Man, and

Cybernetics, 34(2):138–153.

Murphy, R. R. (2014). Disaster Robotics. The MIT Press.

Murphy, R. R., Tadokoro, S., Nardi, D., Jacoff, A., Fiorini,

P., Choset, H., and Erkmen, A. M. (2008). Search and

Rescue Robotics, pages 1151–1173. Springer, Berlin.

Neerincx, M. (2003). Cognitive task load analysis:

allocating tasks and designing support, chapter 13,

pages 283–305. Lawrence Erlbaum Associates, NJ.

Riley, J. M. and Endsley, M. R. (2004). The hunt

for situation awareness: Human-robot interaction

in search and rescue. Proceedings of the Human

Factors and Ergonomics Society Annual Meeting,

48(3):693–697.

Rivero, C. R., Hern

andez, I., Ruiz, D., and Corchuelo,

R. (2013). Benchmarking data exchange among

semantic-web ontologies. IEEE Transactions on

Knowledge and Data Engineering, 25(9):1997–2009.

Schlenoff, C. and Messina, E. (2005). A robot ontology

for urban search and rescue. In Proceedings of

the 2005 ACM Workshop on Research in Knowledge

Representation for Autonomous Systems, KRAS ’05,

pages 27–34, New York, NY, USA. ACM.

Sheth, A. P. (1999). Changing Focus on Interoperability in

Information Systems: From System, Syntax, Structure

to Semantics, chapter 2, pages 5–29. Springer US,

Boston, MA.

Simperl, E. P. B. and Tempich, C. (2006). Ontology

engineering: A reality check. In Meersman, R. and

Tari, Z., editors, On the Move to Meaningful Internet

Systems, volume 4275 of Lecture Notes in Computer

Science, pages 836–854, Berlin, Heidelberg. Springer.

Tenorth, M., Perzylo, A. C., Lafrenz, R., and Beetz,

M. (2013). Representation and exchange of

knowledge about actions, objects, and environments

in the roboearth framework. IEEE Transactions on

Automation Science and Engineering, 10(3):643–651.

Xu, W. and Zlatanova, S. (2007). Ontologies for Disaster

Management Response, pages 185–200. Springer,

Berlin, Heidelberg.

Ontology Design for Task Allocation and Management in Urban Search and Rescue Missions

629