ArchivIA - Archivio istituzionale dell'Universita' di Catania >
Tesi di dottorato >
Area 01 - Scienze matematiche e informatiche >
Utilizza questo identificativo per citare o creare un link a questo documento:
|Autori: ||Giummarra, Andrea|
|Titolo: ||Uwb-based real-time location system performances
for cow identification and localisation and cow's
location data analysis and management in free-stall barns|
|Abstract: ||The main objective of this study was to evaluate the localisation and identification performances of a Real-Time Location System (RTLS) based on
Ultra Wide Band (UWB) technology within a free-stall barn which represents
a particularly harsh environment for the functioning of this kind of
system. Each dairy cow was equipped with an active tag applied to one ear.
A video-recording system was installed in the barn to perform the assessment
of the RTLS. Top-view camera images of the area of the barn under study
were rectified and synchronised with the RTLS. Each position of the cow
computed by the RTLS was validated by performing cow visual recognition
on the camera images.
To perform this validation a software specifically designed for the purpose
was utilized. It is a quick, accurate, automatic and interactive tool which
includes selection and control tabs for data management, visualization and
labelling of the images with the aim of the computation of the tag true
Localisation and identification performances of the RTLS were assessed
by applying an outlier data cleaning technique to tag localisation errors and
using precision and sensitivity indices. Trade-off between these performances
was found through the computation of three performance metrics.
The results showed that, in the environmental conditions of the barn, the
RTLS produced errors which were comparable to the errors declared by the
RTLS producer for the fixed reference tag whereas localisation errors related
to the tags in movement were higher and, in detail, a mean error of 0.56 m
and an error at the 90th percentile of 1.03 m were obtained.
Outlier data cleaning produced significant improvements of the results by
reducing the average localisation error of about 0.046 m for the cows tags and
made data distribution more homogeneous. Trade-off of RTLS performances
yielded an average localisation error equal to about 0.52 m with a position
accuracy of 98% for cows tag and an error of about 0.11 m with a nearly
100% accuracy for the reference tag.
RTLS performances in the considered environment proved to be generally
not dependent on cow behaviour, as it is observed for other systems, and that
RTLS is suitable to determine the occupancy level of the different functional
areas of the barn, compute cow behavioural indices, and track each animal
of the herd.
An automatic and real-time software tool for the visual analysis of the
cows location data in free-stall barns acquired by the RTLS was designed and
developed. A visual representation used for visualizing two dimensional data
of each cow using colours was implemented as a functionality of this tool.
The different colour and colour intensity denoted the difference in sample
density at a location. Measurement of the instant speed of each cow over the
time, represented through an interactive graph, was another functionality
implemented in this software.
The results obtained by a use case of this software tool, which was carried
out in order to acquire useful informations related to the occurrence of estrus,
showed that a pattern, related to the behaviour of the cow analyzed, can be
identified when the state of estrus occurs. Moreover, since it was designed
to monitor cow behavior in real time, it offers the ability, by adding new
control modules, to notify any inconvenience through alert messages as a
result of changes in dairy cow behaviour and, therefore, it is possible to alert
the farmer in real time.|
|In||Area 01 - Scienze matematiche e informatiche|
|GMMNDR85B14M088A-TesiAG (2).pdf||Tesi AG||16,87 MB||Adobe PDF||Visualizza/apri
Tutti i documenti archiviati in ArchivIA sono protetti da copyright. Tutti i diritti riservati.