1
Smart Computing and Sensing Technologies for
Animal Welfare: A Systematic Review
Admela Jukan
+
, Xavi Masip-Bruin
++
and Nina Amla
+++
Technische Universit
¨
at Carolo-Wilhelmina zu Braunschweig, Germany
+
Universitat Polit
`
ecnica de Catalunya (UPC), Spain
++
National Science Foundation, USA
+++
Abstract—Animals play a profoundly important and intricate
role in our lives today. Dogs have been human companions for
thousands of years, but they now work closely with us to assist
the disabled, and in combat and search and rescue situations.
Farm animals are a critical part of the global food supply
chain, and there is increasing consumer interest in organically
fed and humanely raised livestock, and how it impacts our
health and environmental footprint. Wild animals are threatened
with extinction by human induced factors, and shrinking and
compromised habitat. This review sets the goal to systematically
survey the existing literature in smart computing and sensing
technologies for domestic, farm and wild animal welfare. We
use the notion of animal welfare in broad terms, to review
the technologies for assessing whether animals are healthy, free
of pain and suffering, and also positively stimulated in their
environment. Also the notion of smart computing and sensing is
used in broad terms, to refer to computing and sensing systems
that are not isolated but interconnected with communication
networks, and capable of remote data collection, processing,
exchange and analysis. We review smart technologies for domestic
animals, indoor and outdoor animal farming, as well as animals
in the wild and zoos. The findings of this review are expected to
motivate future research and contribute to data, information and
communication management as well as policy for animal welfare.
Index Terms—Smart sensing, smart computing, smart agri-
culture, animal welfare, animal-computer interaction, wearable
computing
I. INTRODUCTION
S
mart computing and sensing have become common terms
to describe next generation computing, communication
and sensing technologies and systems, with a broad range of
Internet and cloud-based applications and connectivity modi,
including combination of various paradigms. The usage of the
term smart may vary, but is typically a networked system
connecting physical devices with computing systems for data
collection, processing, exchange and analysis, - much unlike
stand-alone and isolated systems of the past. Examples of
the basic components of smart computing and sensing today
are networked devices for wearable computing, wireless and
wireline sensor and next generation cellular networks, energy
efficient computing and sensing systems, and big-data process-
ing and visualization. These smart technologies are creating,
and expected to continue making huge societal and economic
benefits in many non-traditional areas.
One of the sectors expected to benefit from the smart
computing and technologies is animal welfare. We use the
notion of animal welfare in broad terms, in consideration
of animal basic needs, health, whether animals are free of
pain and suffering, and also positively stimulated in their
environment, – all for which smart sensing and computing
technologies can play a significant role. Consider the case
of livestock agriculture. While there is no universal United
Nation’s declaration on animal welfare aspects in the context
of sustainable development or best practices recommended
for responsible investments in agriculture, it is rather explicit
that animals are an essential part of sustainable agriculture,
food safety, human health and environmental protection. Since
significant investments are to be made in new technologies for
agriculture, there is no doubt that the same technologies can be
used to monitor and control animal welfare, regionally, state-
wise, and one day, even globally. For instance, the US Animal
welfare law called Twenty-Eight Hour Law that regulates the
maximum length of interstate transportation of animals raised
for food
1
, can easily be supported within smart transportation
systems today, whereby vehicles are connected to the cloud.
Advanced tracking and monitoring technologies have al-
ready been used for pets and wild animals. Under Article 4 of
1987 European Convention for the Protection of Pet Animals,
pet owners must provide their pets with sufficient food, water,
and exercise; today, the latter can be easily monitored by GPS-
and cellular network-based animal trackers. Furthermore, a
new branch of computer science, called Animal-Computer
Interface (ACI) has evolved focusing on improving the human-
animal communications and enabling the so-called animal wel-
fare science. For wild animals, on the other hand, emphasis has
been on systems that non-intrusively monitor their behavior, on
monitoring environmental changes that lead to behavioral and
species-specific issues, as well as co-existence of humans and
wild animals, be it through prevention of road-side accidents,
or preventing illegal hunting of endangered species. To record,
share and analyze biomedical data of animals globally, the
large volume of data produced can only be handled by systems
deeply rooted in today’s notion of clouds, high-end computing
and real-time data transmission. As it is, there is a high
synergetic momentum to revisit smart computing and sensing
systems for domestic and wild animals, foster their further
advances, and pioneer the developments in the area of smart
systems for farm animal welfare, all under a joint framework.
1
United States Department of Agriculture United States Department of
Agriculture National Agricultural Library, Text of the Twenty-Eight Hour Law
(transportation of animals), amended 1994
arXiv:1609.00627v1 [cs.CY] 1 Sep 2016
2
This paper sets the goal to review literature on smart
computing and sensing technologies in the domain of animal
welfare including domestic, farm and wild animals. The review
provides a categorization of smart systems implemented or
discussed in research communities in last decade, or coinciding
with the evolution of the Internet, cloud computing and smart
sensing. While the overall goal of the paper is to improve
animal welfare, and foster technology and science innovation,
the focus in this survey is strictly on categorization of related
smart technologies, providing the basis to manage the informa-
tion and collect data, and helping improve knowledge sharing.
Our findings show that innovative smart technologies appear to
be a promising and economically sustainable option to ensure
animal welfare. The challenges and opportunities discussed
show the richness of the space for technology innovation, and
wide societal benefits, including opportunities to build eco-
nomically sustainable animal welfare systems. While policy
considerations are outside the scope here, relevant stakeholders
may use our findings to facilitate the policy initiative, or
stimulate ethical, economics or legal discussions.
The rest of the paper is organized as follows. Section II
defines the scope of the review, and summarizes the main
criteria used. Section III is dedicated to the technologies and
systems for pets, and companion animals, generally referred
to as domestic animals. Section IV reviews the area of smart
animal farming. Section V is dedicated to smart sensing and
computing systems in the wilderness. Section VI presents
the main findings from the review and discusses briefly the
research opportunities. Section VII concludes the paper and
provides recommendations for further research.
II. SCOPE AND CRITERIA FOR THE REVIEW
Based on standard review methods used in other disciplines
[1]–[3], we follow three steps: planning, conducting, and the
reporting the results of the review (focus of this paper). This
section briefly outlines the first two phases, as the rationale
for the resulting third phase.
A. Planning the Review
This survey uses the definition in [4] where the animal wel-
fare subject was studied from a cross-disciplinary perspective
of the so-called animal welfare science and animal-computer
interaction in particular. For our purposes, the relevant part
of animal welfare science is the technology that can produce,
process and use data to allow research and policy making in
the criteria relevant to animal welfare, such as: (i) animals
living without pain, (ii) control of species-adequate living
environment and (iii) positively stimulated activities and social
interactions of animals, both with humans and other animals.
As such, our review does not go into specific aspects of
ethics, animal rights and laws. The term animal welfare is
strictly reviewed in the context of smart sensing and computing
technologies.
Whilst the subject of intense research in a number of
different application areas, in this survey we focus only on
smart technologies that involve animals. We paid attention
to the accessibility and reproducibility of the studies con-
ducted, in the context of specific technology or devices used.
The survey excludes the following sources (i) commercial
products and the associated white papers; (ii) opinion, op-
ed, journalistic articles, books and book chapters, position
papers, (iii) technological innovation of individual components
potentially applicable, but outside the area of animal welfare,
(iv) technological studies in the area of anthropomorphism,
concerned with human-centric attribution of animal welfare
features, and (v) any technology and systems built for the
solely purpose to address issues of animal law, rights or ethics.
Note that a significant amount of research work in robotics
and virtual reality has been dedicated to creating animal-like
robots, serving in similar roles as live animals. For instance,
University of Sidney has recently reported the deployment
of a SwagBot
2
, a robot used for herding and monitoring
cattle on a farm. Also, robotic dogs have been developed
and used as service animals [5]. Similarly, canines in virtual
environments seem like a promising alternative as companion
or for therapeutic purposes [6]. This review does not cover
these and similar efforts in robotics and virtual reality, and
include only research where live animals are considered.
B. Conducting the Review
Figure 1 illustrates the review conducted according to the
perspectives we have taken and categories considered. From
the technology perspective, we focus on four main categories
of the work reported: communication, health, monitoring and
environment. Communication refers to the systems that enable
humans to communicate with animals. Important aspects of
communication are capturing the type of data exchanged, and
storing and using this data for analysis and processing. The
category Health includes aspects of both animal and human
health. This could include smart systems to monitor animal
health, as well as technologies that employ animals to assist
disabled people or other therapeutic treatments. Monitoring
relates to (remote) monitoring of the animal behavior. The
category Environment relates to monitoring the indoor and
outdoor environment of the animals.
We define three major categories of animals: domestic, farm
and wild. The category Domestic animals refers domestic
pets, service animals and working animals. We define service
animals as those trained to help a disabled individual, and
working animals as those trained to help society at large like
military and search and rescue. In this category, we review the
systems intended for use on an individual animal (be it dog,
cat, cow, or horse). The category Farm animals refers to a
group of animals reared for the animal products, and generally
housed together in a farming facility. The actual species of
the animal is unimportant, but the technology designed for
a group of animals designated for human food production
(dairy or meat) or commercial goods (wool). The category
of Wild animals refers to animals in their natural habitats, or
in confinement in zoos or sanctuaries.
2
New Scientists (online), Cattle-herding robot Swagbot makes debut on
Australian farms, July 2016
3
The review quantifies the work done in each category,
and identifies current relationships between individual sub-
categories. While Figure 1 does not show the relationship be-
tween individual categories and subcategories, these relation-
ships can be important. The notable absence of experimental
and research papers in one of the categories could indicate the
need for future research in that space, or that the specific area
or application is not important.
III. DOMESTIC ANIMALS: ONE AT A TIME
The focus in this section is on domestic animals, where the
distinguishing factor is that they are treated individually, and
not as a group. Following the classification proposed in Figure
1, we review domestic animals in three main categories of
applications: (A) human-animal communication, (B) tracking,
behavioral monitoring and animal health, and (C) service dogs
and working dogs (Table I). These are summarized in Table
II.
A. Human-animal communication
The human curiosity for communicating with their animal
companions, primarily dogs and cats, is probably as old as
the history of domestic animals. Today’s technology make it
possible to articulate this communication through a focused
and distinctive subfield of computer science called Animal
Computer Interface (ACI) [33]. ACI and animal welfare are
naturally aligned, as discussed in [8], considering the cross-
disciplinary collaboration that it can offer. For instance, it
is well established in the ACI community that the design
of interfaces for dogs should involve technology developed
solely for their use and designed based on species appropriate
needs [34]. In one of the pioneering works, the authors of
[9] propose a cybernetics system that transfers human contact
through the Internet to a chicken, for its therapeutic effects on
both chicken and humans. The system transfers the chicken’s
motion to a physical doll on a XY-axis positioning table or
as a real-time 3D live view of the chicken. A significant part
of the research efforts in the ACI area focuses on positive
stimulation environments for pets, as playing is seen as one
of the most natural and inherent behaviors of animals. In [7],
digital games were proposed for cats in a multimodal virtual
environment deploying kinetic sensors indoors.
A prototype for human-dog communication based on a
smartphone attached to the dog, including the communication
about various senses such as smelling, hearing, touching,
vibration, and testing food, was demonstrated in [14]. This
extensive portfolio of communication with the animal can
be used to train service and working animals, in addition to
improving human-animal interaction. It was found in [35] that
even a simple GPS enabled collar can improve human-animal
interaction. The case studies in [36] on dog owners’ needs and
expectations towards communication technologies revealed the
limitations in usability of the current systems and applications.
A specialized social media platform for pets was proposed
in [31], where the pet’s activity is automatically monitored
through RFID activity tags they carry, and automatically
posted on social networks. [32] proposed to extend social
media to non-human species. A pet video chat system based
on Skype was proposed in [27].
B. Tracking and monitoring of human’s best friends
The interpretation of dogs’ postures has been subject of
significant research in part to better understand their behavior
in natural environments, and in part to analyze their eating
and sleeping patterns, based on wearable activity recogni-
tion systems, such as in [10]. The authors of [11] used
accelerometer and gyroscope data provided by a wireless
sensing system deployed on a dog’s vest. The system uses
machine learning algorithms to interpret the dog’s postures,
like sitting, standing, lying down, standing on two legs and
eating off the ground, as well as dynamic activities, like
walking, climbing stairs and walking down a ramp. Similarly,
[12] proposed algorithms for the recognition of dogs’ postures,
and also for non-domesticated terrestrial mammals in general
[25], based on studies with an Eurasian badger. A dog-to-
handler communication system in [26] enables bidirectional
communication with dogs who carry sensors and GPS, and can
activate signal triggers, and handlers sending vibration signals
to the dog. Finally, a wireless health monitoring system for
dogs was proposed in [15] to gather and analyze the health
data through a wearable jacket.
C. Service dogs and working dogs
In a typical scenario, one service dog is dedicated to
one person with chronic health conditions, such as visual
or physical impairment, epilepsy or diabetes. A user-friendly
canine alarm system for service dogs based on a pull-off
trigger monitored by a Raspberry Pi was proposed in [30].
The authors of [17] propose communication systems with
audio and vibrotactile feedback for blind people to monitor
their guide service dogs and to interpret their dogs’ feelings
and body language. The work presented in [16] evaluates
dog interfaces for alarm systems, which allow diabetes alert
dogs to remotely call for help when their dedicated human
companion falls unconscious. The paper discusses the needs
of individual dogs when designing such interfaces. Similarly,
[37] argues that guide dogs, when off work, are just pets that
have basic needs like feeding, grooming, attention, playing
and free running. Therefore, as this paper suggests, research
on accessible dog toys utilizing sensor technologies for guide
dogs, to improve their welfare, is an important future direction.
A pilot study based on the use of activity trackers for the
assessment of service dogs, which show how suitable an
individual dog is for a specific work, was conducted in [38].
Service dogs can also be used in human health care domain
as therapy dogs. Paper [39] surveys the Medline, PsychInfo
and CINAHL databases for research papers on the effect of
animal-assisted therapy for dementia. Animal-assisted therapy
appears to be beneficial for people with dementia, and carries
potential for technological innovation in animal computer
interfaces for therapy dogs. Notably also, it was demonstrated
in [18] how a cancer detection dog can put different pressure
on the positive and negative cancer samples while sniffing
4
Smart& Computing&an d& Sensing&Applications& fo r& Animal&Welfar e&
Commu nication Health Monitoring Environment
Data&
Excha nged
Animal&
Health
Animal&
Computer&
Interface& (ACI)
Behavior
Human&
Health
Control Monitoring Control
Animal& World&&&&&&aaaaaaaa
Domestic Farm Wild
Pets
Service
Anima ls
Working
Anima ls
For&animal&
products
For&human&
food
Endang ered&
species
Scientific&
research
Anima ls&in&
confinement
Policy,&Ethics,& Envir onmental& Sustainability
Tech no l o g y & In n ov ati o n& an d & S ci en ti fic& D i s co ver y
Fig. 1: Classification of animal welfare attributes reviewed based on animal centric and technology attributes
them, and this can be recognized by monitoring this pressure
with sensors.
Unlike service dogs, working dogs are typically dedicated
to a task, rather than to individual humans. They have been
used in search and rescue, and military combat situations, and
are typically equipped with sophisticated wearable devices.
The smart computing and sensing augmentation of Urban
Search and Rescue (USAR) dogs has been proposed in various
combinations. These could be wireless cameras mounted on
the dog’s shoulders as proposed in [19], or a combination
of cameras, microphones, speakers, GPS, and networks, as
proposed in [20]. The ongoing development of a CAT named
telepresence system for USAR dogs is reported in [28]. The
authors of [21] propose the detection of continuous barking,
derived from audio and body motions of USAR dogs, signaling
the localization of victims searched. Likewise [22] proposes
to transmit the pose of USAR dogs every 50ms through an ad-
hoc mesh network, to interpret the dog’s intention and predict
search and rescue success. A motion sensor on a dog’s collar
for communication via the use of head gestures is proposed
in [13]. Of particular concern is the monitoring of health of
USAR dogs in extreme conditions, such as weather (heat),
for which [23] propose health monitoring for USAR dogs. In
another scenario, USAR dogs are used in combination with
robots, and are thus protected without compromising search
and rescue missions. In [24], USAR dogs carry snake robots
into areas that are inaccessible for their human handlers and
too dangerous, or too narrow for dogs.
IV. FARM ANIMALS: MANAGING ANIMAL GROUPS
In contrast to domestic animals, in a typical farm setting,
animals are not identified individually with ID cards, chips,
and names, and their welfare is managed in the context of a
group. Unlike companion and working animals, farm animals
are raised for the commercial utility of the products they
can deliver: eggs, dairy, meat, leather, etc. Economic factors
involved in deploying smart systems play an important role in
this context. This section reviews research on smart technolo-
gies for farm animals as a group, focusing on applications
to cows, pigs, chickens, rabbits and sheep. We organize the
review according to the two main habitation categories for
farm animals: indoor and outdoor. Indoor animal farming is the
5
TABLE I: Classification of applications and data exchanged in smart compute and sensing systems for domestic animals
Application / Papers
Data Exchanged Wearable
Non-wearable
Human-animal communication [7]–[9]
human contact exchange, location,
posture, heptic interfaces, orientation
vibra-tactile actuators
mobile computer-pet jacket
kinetic sensors
3D visuzalization
Tracking and
behavior monitoring [10]–[14]
postures (sitting, standing, lying down),
sleeping and eating patterns
walking, climbing stairs,
sending vibration or audio signals to dog
wireless accelerometer and gyroscope
motion sensor
GPS sensors, RFID tags
speakers on the harness
smart phones
mobile networks
social network
Animal health [15]
heart rate (HR)
heart rate variability (HRV)
respiratory rate
vital signs
electrocardiogram (ECG) electrods
photoplethysmogram (PPG) sensors
inertial measurement units (IMU)
optical fibers and lightguides
computational node
Human health
(service and working dogs) [16]–[18]
remote call for help
barking
vibrotactile feedback to humans
wearable sensors, conductive
polymer potentiometer sensor
computational node
Search and rescue
(working dogs) [19]–[24]
continuous barking detection,
posture detection
animal vital signs detection
wearable wireless cameras
microphone, speakers, GPS
gas sensors
EKG and PPG sensors
robots
wireless networks
TABLE II: Key system technologies proposed for domestic animals
Key technology Discussed in papers
Accelerometers [11] [19] [15] [10] [21] [25]
Gyroscopes [11] [21]
Vibrotactile response [17] [26] [23]
Video [24] [20] [23] [21] [27] [28]
Audio [17] [20] [23] [27]
Animal input interface [16] [26] [18] [29] [30]
Animal health sensors [17] [15] [23]
Microcontroller / SoC [11] [15] [23] [10] [21] [29] [28]
Wireless Mesh / WiFi [22] [15] [23] [21]
Bluetooth [11] [19] [15]
GPS [20] [26] [23] [21]
RFID [31]
Mobile application [14] [15]
Networked service [9] [31] [32]
Behavior detection systems [11] [22] [23] [10] [13] [12] [18] [21]
Robotics [24]
3D live virtualization, avatars [9]
most common kind of farming, with largest amount of work
reported. Table III summarizes applications, research papers
and also research questions addressed in indoor farming, cat-
egorized according to the species. Outdoor farming practices,
generally viewed as a more natural setting for animals, have
also been subject of research. This is summarized in Table IV
and described in Section IV-B. We summarize the technologies
reviewed for farm animals in Table V.
A. Indoor farm animals
Farm animals raised indoors present interesting case studies
for smart technologies, integrating smart building and energy
innovations with animal welfare, creating a coordinated smart
ecosystem. The work reviewed presents isolated parts of that
vision, often motivated by the economic factors of animal
farming, and reflected through animal health, and consequently
the quality of resulting animal products. A fair portion of work
surveyed focuses on activity monitoring and indoor tracking,
directly applicable to the animal’s ability to move indoors.
This is a critical welfare factor, since in most cases these
animals remain in that setting for their entire lifetime. Using
smart technology for more efficient animal farming, both for
economic and welfare reasons, has surprisingly received far
less attention than agricultural farming.
Paper [40] monitors the state-of-health of cattle remotely,
and develops a veterinary telemedicine infrastructure that
includes wearable sensors and a Bluetooth system. Paper [41]
proposes a system that uses magnetometer and accelerometer
technology to monitor heart rate, and activity level in cattle.
The control sensors are equipped with a low power wire-
less routing protocol, which presents engineering challenges.
Cow’s estrus, heat stress and onset of calving have been the fo-
cus in [42] and [43]. The proposed systems use ZigBee based
wireless sensor network to detect the body temperature and
movement. Another effort [44] focuses on detecting lameness
in cows using camera sensors in real time to detect the curve
formation by the head position and back posture. Paper [45]
deploys a sensing climate control system for indoor cattle
farming to improve the comfort level of animals and detect
disease. The focus in [46] is on finding the location of indoor