Heriot-Watt University
ROS: The Research Output Service. Heriot-Watt University EdinburghNot a member yet
4689 research outputs found
Sort by
Unfolding neurorobotics as a tool to investigate regular and disrupted neural activities using humanoid robots
No full textThis Ph.D. thesis explores the field of neurorobotics and its application in understanding the coordination of movements under regular and disrupted neural activities, specifically in the context of Parkinson’s disease. Parkinsonism is a neurodegenerative disorder characterized by motor impairments, including bradykinesia,
rigidity, and tremors. The aim of this research is to leverage neurorobotics as a tool
to study and comprehend the human motor control in a controlled environment and
to support the investigation of neurological disorders. The thesis begins with a short
description of biological aspects of the human motor control, and a short review of
neurorobotics, focusing on the neural circuits and the processes involved in motor
control. The review establishes a foundation for the subsequent chapters, which
explore the development and implementation of our proposed neurorobotics models.
Our first contribution involve the embodiment of an autonomous neural systems in
a humanoid robot. We embedded a composite model of the basal ganglia, thalamus
and cortex system in a real humanoid robot to perform movements based on sensory information. In our experiments, the robot was able to sense the environment,
process the information and act accordingly to the behavioural task. Experimental evaluations were then conducted using the neurorobot under both healthy and
parkinsonian conditions. Then, we built a more sophisticated neurorobotics model
to coordinate movements in the robot’s arm. We designed a computational model of
the cerebellum and integrated it to our basal ganglia, thalamus and cortex system.
Our integrated model uses biological features to move the arm of the robot. The
results demonstrate the ability of the neurorobotics model to execute motor commands under different neural conditions. Some motor impairments seen in Parkinson’s disease could be observed which shed light on the underlying mechanisms of
Parkinson’s disease. The implications of this research extend beyond the field of
neurorobotics. By gaining a deeper understanding of the motor disturbances, this
work may contribute to the development of more effective therapeutic strategies and
interventions. The neurorobotic platform serves as a valuable tool for exploring and
testing potential treatments, such as deep brain stimulation, in a controlled and
adjustable setting. In conclusion, this Ph.D. thesis utilizes neurorobotics as a novel
approach to study the human motor control in different scenarios. The research
contributes to our understanding of the brain and the mechanisms of Parkinson’s
disease which may paves the way for advancements in the diagnosis and treatments.
Additionally, we support the usage of neurorobotics models as a means to minimize
the reliance on animal experimentation in medical research
Advances in precision optical metrology : from SPAD arrays to non-linear optics
No full textThis thesis details several methods and applications for measuring the arrival time
of optical signals. The first of these methods measures the arrival time of single photons using electronic signals via single-photon avalanche detector (SPAD)
technology. These detectors have been implemented in light detection and ranging
(LiDAR) experiments, with pico-second timing resolutions allowing depth images to
be reconstructed to millimeter precision. Another method uses optical signals via
quantum interference and non-linear optics to measure photon arrival time, with
attosecond precision measurements of the temporal separation of pulses observed.
Within this body of work, there are details of four optical experiments. The first
experiment is in the field of Light-in-flight (LIF) imaging, which is the measurement
and reconstruction of light’s path as it moves and interacts with objects. While
it is well known that relativistic effects can result in apparent velocities that differ
significantly from the speed of light, it is less well known that Rayleigh scattering
and the effects of imaging optics can lead to observed intensities changing by several
orders of magnitude along light’s path. We develop a model that enables us to
correct for all of these effects, and from this we reconstruct the true intensity-corrected optical path of a laser pulse in four dimensions (three spatial dimensions
and time) as it travels in air.
The second experiment is in the field of non-linear optics, and proposes a method
for the precise measurement of a target depth, with applications in biophysics. By
exploiting the high sensitivity of an autocorrelator’s dependency on path length,
we propose a technique that achieves ≈30 nm depth resolution for each pixel in 30
seconds. Our method images up-converted pulses from a non-linear crystal using
an sCMOS camera to convert the intensity recorded by each pixel to a delay. By
utilising statistical estimation theory and using the data from a set of 32×32 pixels,
the standard error (SE) of the detected delay falls below 1 nm after 30 seconds.
In the third experiment, we explore the group velocity of spatially structured optical modes, which exhibit a group velocity lower than c, resulting in a measurable
temporal delay with respect to plane waves. Here, we develop a technique to image
this temporal delay and measure it across a set of optical modes. An inevitable consequence of spatially varying delay is temporal broadening of the mode. As such,
for a focused Gaussian, we observe an ≈ 1% increase in the temporal profile, corresponding to a narrowing of the optical spectrum by ≈ 0.03 nm. This work shows
that imaging is essential to fully understanding the changes to the group velocity
for structured modes.
In the final experiment, we explore the potential use of a GHz rate laser in quantum
optics. The generation of single photons through spontaneous parametric down-conversion (SPDC) has applications in many quantum optics experiments. Applications such as satellite quantum key distribution (QKD) require a compact, high
generation rate single photon source. Here we propose the use of a compact, cheap
GHz diode pumped three-element Ker-lens-modelocked Ti:sapphire laser as a single
photon source. We verify the presence of single photons produced via SPDC using
Hong-Ou-Mandel (HOM) interferometry and observe a dip in coincidence counts
with a visibility of 81.8%
Development of biomass based activated carbons for biogas upgrade
Full text linkThe direct combustion of biogas has the problem of corrosion in wet conditions, low
heating value and limited applications. The removal of CO2 to produce biomethane
(biogas upgrade), addresses these diverse challenges. Among the various techniques
available to remove the CO2, adsorption-based systems using activated carbon (AC)
obtained from biomass precursors are gaining importance due to their cost effectiveness
and better performance (superior capture capacity and CO2/CH4 selectivity).
This doctoral thesis presents a comprehensive investigation into the development and
application of biomass-derived (spruce sawdust (SD), miscanthus straw (MS), wheat
straw (WS), and sewage sludge (SS)) physical and chemical activated carbons for CO2
capture from biogas. The research delves into the detailed optimisation of the
experimental conditions, systematic characterisations (TGA, Fourier transform infrared
(FTIR) analysis, CHN analysis, and adsorption isotherms), and breakthrough studies.
The research concludes with the identification of an optimal char sample exhibiting cyclic
stability and remarkable CO2 capture capacity of 1.98 mmol/g under 90 % CO2 stream.
Additionally, physically, and chemically activated carbons are highlighting varying
activation conditions and their impact on textural properties and CO2 capture capabilities.
The optimum physical and chemical activated carbons have high CO2 capture capacities
(2.42 mmol/g by ACSD (860-1-30) and 3.64 mmol/g by ACMS (700-10-90-1.5) at 25 °
C.
The breakthrough (dynamic) studies of the optimum samples in a fixed-bed reactor under
various biogas compositions (30/70, 50/50, and 70/30 vol % of CO2/CH4) uncovers rapid
adsorption kinetics and breakthrough profiles (S shaped), highlighting the preferential
adsorption of CO2 over CH4. The influence of gas concentrations on separation efficiency
is carefully analysed, highlighting exceptional performance (selectivity above 3 by all the
optimum samples under 30 % CO2 stream and a maximum selectivity of 3.76 by ACSD
(860-10-30) under 50 % CO2 stream at 25 ° C and atmospheric pressure. The CSD (700-
10-60) has better separation performance (selectivity of 3.59 under 30 % CO2 stream) as
compared to many chars and physically activated carbons.
In summary, this research not only expands the understanding of biomass-derived
activated carbons but also provides valuable insights for their practical application in CO2
capture. The findings contribute to current efforts in developing sustainable materials to
address environmental challenges, particularly in the domain of greenhouse gas
mitigation and sustainable alternative fuels
Sub-pixel microscanning of a single-photon detector array based LiDAR system for improved image reconstruction
No full textLong-range light detection and ranging (LiDAR) imaging using the time-correlated single-photon counting (TCSPC) technique is a well established method for obtaining high-resolution
depth profiles at long standoff distances. The high sensitivity and excellent surface-to-surface
resolution provided by the TCSPC technique allow for high quality depth profiles to be obtained
from photon time-of-flight (ToF) information while using low average optical power levels.
Current single-photon LiDAR systems operating in the short-wave infrared (SWIR) benefit from
a reduced solar background at these wavelengths, improved atmospheric transmission, and a less
restricted eye-safety threshold allowing for safe use of higher laser powers compared to systems
operating at shorter wavelengths.
However, many SWIR single-photon LiDAR systems are limited by the low pixel format
of the detector employed. Currently, the largest single-photon detectors operating in the SWIR
regime have are on the order of 1000s of pixels, producing images with lower spatial resolution
when compared to CMOS-based single photon detector arrays with typically greater than 100,000
pixels operating in the visible and near infrared regions. Currently, larger format SWIR single-photon detector arrays are prohibitively expensive to develop, hence microscanning offers an
approach to improved image reconstruction using lower format focal plane arrays. The selected
microscanning approach in this Thesis involves moving the image sensor to many locations in
the image plane, and acquiring a unique image at each location. Each of these lower format
images are then combined together to form a single, more detailed image.
The work presented in this Thesis will describe the application of the microscanning technique
to a single-photon LiDAR system operating in the SWIR. This system is comprised of a 32×32
InGaAs/InP single-photon avalanche diode (SPAD) detector array, and a pulsed fibre laser source
with the TCSPC technique used to provide ToF information of remote targets. Results presented
were obtained from long-range imaging experiments at two standoff distances of 325 m and
1.4 km. A number of challenging imaging scenarios were investigated at long range, including
detailed and complex scenes, and the reconstruction of moving targets
A nonlinear elasticity model in computer vision
Full text linkIn this thesis, we formulate a nonlinear elasticity based model for the comparison and registration of pairs of images in a computer vision context. We
establish the existence of minimisers of an integral functional intended to
capture differences between images, and hence obtain ‘optimal’ transformations for this purpose. We impose several desirable conditions on the elastic
potential to guarantee matching of images related in particularly straightforward ways i.e magnifications, rotations, shears. We establish variants of
the central existence theorem for particular subclasses of image registration
and identification problems. In the case of linearly related images, we show
that a particular form of elastic potential is required, and modify the approach by adding second derivative terms to preserve existence of minimisers
in this case. We provide a constructive proof allowing smooth extensions of
(possibly infinite valued) proper convex lower semicontinuous functions onto
a halfspace. This clarifies a technicality on the proper formulation of poly-convexity in nonlinear elasticity as the determinant approaches 0, allowing a
weakening of hypotheses.Engineering and Physical Sciences Research Council (EPSRC) grant EP/L016508/
Bayesian reconstruction and regression with multivariate graph signals
Full text linkGraph Signal Processing (GSP) is a rapidly evolving field that combines ideas from spectral graph theory and classical signal processing to analyse and manipulate data residing
on an irregular domain. In this thesis, we contribute several advancements to GSP theory, in particular, with regard to Bayesian reconstruction and regression techniques for
multivariate graph signals. The first topic we consider is the reconstruction of signals
existing on two-dimensional Cartesian product graphs, in the presence of noise and arbitrary missing data. Using numerical methods and the properties of the Kronecker
product, we derive two efficient algorithms for computing the posterior mean and show
how the optimal choice of technique depends on the model hyperparameters and sparsity
of the input data. We then build on this by applying similar algorithms to solve several
multivariate graph signal regression models. In particular, we generalise prior work on
Kernel Graph Regression (KGR) and Regression with Network Cohesion (RNC), which
are relevant when the explanatory variables are exogenous and endogenous respectively,
by allowing for arbitrary patterns of missing data in the input signal. Following this,
we adapt the reconstruction and regression methods developed prior in the thesis to
the Multiway Graph Signal Processing (MWGSP) paradigm. MWGSP is an emerging
sub-field that focuses on tensor-valued graph signals, where each axis is described by a
unique graph topology. In order to help write effective and efficient MWGSP algorithms,
we also present the PyKronecker library which creates an abstracted API for manipulating high-dimensional Kronecker-structured matrices. The next topic we consider is
techniques for computing the posterior covariance of our models. First, we propose
several algorithms for estimating the marginal posterior variance and compare them to
other alternative standard techniques. Combined with an active learning strategy, we
demonstrate that our procedure can generate superior estimates, with R2 > 0.95. We
also derive an efficient algorithm for sampling directly from the posterior whilst avoiding computationally expensive MCMC-based approaches, using a technique known as
perturbation optimisation. Finally, we develop new models that generalise our previous reconstruction and regression models to accommodate binary and categorical tensor
graph signals. Each topic in this thesis is also accompanied by a real-world case study
to corroborate the utility of the methods or demonstrate their theoretical properties
An investigation into the social, political and economic barriers to the adoption of a Mass Rapid Transit system in Malta. A study based on the Extended Theory of Planned Behavior and Behavioural Economics
Full text linkCongestion in Malta is difficult to alleviate. In a country which has a land mass of just
316km2
, in 2020, Malta had a population of 519,562, with 263,352 licensed drivers, 422,576
vehicles on the road, increasing at 65 every day, and with a road network of 2,450km, making
this island the fifth most densely populated nation and the fifth most dense transport network
in the world. The Government of Malta has finally proposed to build an underground Mass
Rapid Transit (MRT) system, a METRO that is estimated to cost €6.2b, take 20 years to
complete and cover 25 stations. For such a system to succeed, it is vital that a significant
number of Maltese licensed drivers opt to switch and use this new public transport.
This study adopted a post-positivist quantitative approach, employed a convenience, non-representative stratified sample, through both a pilot and a main study, employing second
generation statistical methods, specifically Partial Least Squares Structural Equation
Modelling (PLS-SEM), to test the validity, reliability and predictability of a proposed
Extended Theory of Planned Behavior (ETPB) model, that combined the Theory of Planned
Behavior (TPB) and Behavioural Economics (BE). The final resulting model was a
combination of these two theories, as well as the inclusion of the UK Department for
Transport (DfT, 2011) intervention ladder approach. This allowed the application of both
soft-behavioural methods and hard-policy regulations to overcome a number of social,
economic and political barriers, identified in this study that can inhibit adoption and use of
the MRT by Maltese licensed drivers. The final model passes all reliability, validity and
predictability tests, showing distinct path coefficients between the unobservable endogenous
latent variables and the exogenous observable variables.
The study findings confirmed that TPB and BE are coherent interwoven behavioural models,
especially within the social and subjective norms.
This academic study is a first for Malta for a transportation study, which also adopts PLS-SEM as a data analysis process. The study findings identified 12 key social, economic and
political barriers that would deter the adoption and use of MRT by Maltese licensed drivers
and offers 20 recommendations to ensure that prior and leading to the introduction to MRT,
their intent to switch and use an MRT, becomes a behavioural reality
Integration of Industry 4.0 with Lean Management in large German manufacturing firms - a dynamic capabilities perspective
Full text linkThis research is concerned with how large German manufacturing firms can realise the
integration of Lean Management and Industry 4.0. This is relevant for companies that are
engaged in a semi-matured Lean transformation and have not yet realised all key
principles but, at the same time, are confronted with a fourth industrial revolution.
Building on mature but primarily separate research streams of Lean and Industry 4.0, this
study employs an exploratory sequential mixed-methods design to derive a framework
for integrating these important themes of Operations Management and to support firms
that are unwilling to approach a sequential integration of Lean or Industry 4.0 first, as
typically advised by previous research and seeking advice on concurrent integrations.
Based on a review of existing academic literature, the main themes of Operations
Management as the setting for this research, Dynamic Capabilities as the theoretical lens,
and Lean and Industry 4.0 as the subject focus are synthesised, research questions derived
and a conceptual framework formulated.
The research design utilised an exploratory qualitative strand, which derived major
integration themes and 201 potential modes of action through a Thematic Analysis of 16
semi-structured interviews with German subject matter experts. The subsequent
quantitative strand evaluated and prioritised six dimensions and 43 potential modes of
action through a triangulating exploratory survey with 256 subject matter experts from
Germany. Finally, the validating strand utilised a Delphi study with 15 subject matter
experts. It derived a refined and validated framework consisting of 50 items organised in
the six dimensions of ‘initiating’, ‘sensing’, ‘seizing’, ‘transforming’, ‘resources’, and
‘capabilities’ to execute an integration of Lean and Industry 4.0. Consequently, the
findings are influenced by the geographical focus, firm size, theoretical lens of Dynamic
Capabilities, and methodological design, which opens up exciting possibilities for future
research contributions.
This research contributes practically to the field of Operations Management by proposing
executable modes of action and a concurrent pathway as an alternative for firms intending
to integrate Lean and Industry 4.0. It theoretically contributes to advancing Dynamic
Capabilities theory by proposing a novel dimension derived from an application and
concretisation in a new research area
Optimising information gains for increased precision in quantum metrology
Full text linkThe field of quantum metrology encompasses a range of techniques that exploit quantum
phenomena to perform high precision measurements of classical and quantum parameters. In this thesis we look at methods for optimising metrology protocols through finding
measurement approaches yielding greater information. After an introductory chapter out-lining common concepts, this thesis can be broadly divided into two halves: the former
concerning optical interferometry using one- and two-photon states; and the latter exploring quantum sensing with spin systems, with particular consideration of the limits
imposed by using a nitrogen-vacancy centre, a spin defect in diamond, as a quantum
sensor.
In Chapter 2, we look at the Hong-Ou-Mandel interferometer with two-photon input
states. The resulting interference between the two photons provides a mechanism for
estimating an optical path-length difference, and we look to enhance the performance of
the Hong-Ou-Mandel protocol through the inclusion of photon detectors with time and
number-resolving capabilities allowing for higher precision measurements. In Chapter 3,
we compare the performance of the Hong-Ou-Mandel interferometer, which is commonly
understood to boast higher noise resilience, to that of the Mach-Zehnder interferometer,
which can in ideal conditions perform much more precise measurements but at the cost of
a high sensitivity to noise. For the latter, we consider a number of one- and two-photon
inputs, with the combination of single and biphoton interference giving rise to different
optimal inputs for different noise regimes.
Turning to magnetometry with spin systems, in Chapter 4 we explore the implemen tation of three-dimensional quantum gates for use in a Bayesian multiphase estimation
algorithm performed using the ground state triplet of a nitrogen-vacancy centre as our
qutrit, with a specific focus on implementing a Hadamard operation. This can be achieved
through a sequence of magnetic pulses approximating two-dimensional rotations, and we
consider the effects of decoherence in imposing limits on the fidelity of this process. Finally, Chapter 5 outlines an adaptive protocol for estimating the frequency of an AC
magnetic signal. Utilising dynamical decoupling to extend the effective coherence time
of our sensor, we demonstrate adaptive rules capable of far exceeding the precision of an
equivalent non-adaptive estimation
Exploring the impact of sewer-derived airflows on the air-pressure dynamics within building drainage systems
Full text linkThe performance of a building drainage system (BDS) relies on complex internal
airflow and pressure dynamics, governed primarily by the unsteady wastewater
flows from randomly discharging appliances such as WCs, sinks and baths.
Designers attempt to optimise the safety of the system by including pressure
equalisation strategies in the form of ventilation pipes and more active devices such
as pressure attenuators and air admittance valves. Failures within these systems can
compromise water trap seals, allowing hazardous sewer gases to enter buildings.
While these measures can equalise the air pressure within the above ground
drainage system, air coming from the sewer can have an effect on the performance
also. Traditionally, above and below ground drainage systems are designed in
isolation and there is no recognition of the influence of one on the other.
This thesis documents the development of a novel model to represent the impact of
sewer air on the performance characteristics of a BDS, leading to the development
of new conceptual diagrams describing the interaction, that illustrate the correlation
between newly introduced terms, such as; modified entrained air and modified air
pressures, when the system is exposed to both BDS operation and sewer air.
Laboratory experiments were conducted using a full-scale drainage test rig
representing both low rise (3 storey) and high rise (34 storey) buildings that together
provide empirical insights scalable to real-world applications. This approach
bridges the gap between laboratory experiments and real-world dynamics, thereby
enhancing the reliability and applicability of the research findings.
The research confirmed that the airflow and air pressure regime within the vertical
BDS stack is modified by and influenced by the connection to the main sewer in a
manner consistent with an interaction analogous to a fan and system loss curve,
requiring the solution of simultaneous equations describing both. The findings of
this study confirm a direct correlation between pressure fluctuations and building
height when exposed to sewer air
Access Repository Dashboard
Do you manage ROS: The Research Output Service. Heriot-Watt University Edinburgh ? Access insider analytics, issue reports and manage access to outputs from your repository in the CORE Repository Dashboard!