Titel Deelnemers "Korte inhoud" "Real-Time Edge-Sensitive Local Stereo Matching with Iterative Disparity Refinement" "Maarten DUMONT, Patrik GOORTS, Steven MAESEN, Gauthier LAFRUIT, Philippe BEKAERT" "First, we present a novel cost aggregation method for stereo matching that uses two edge-sensitive shape-adaptive support windows per pixel region; one following the horizontal edges in the image, the other the vertical edges. Their combination defines the final aggregation window shape that closely follows all object edges and thereby achieves increased hypothesis confidence. Second, we present a novel iterative disparity refinement process and apply it to the initially estimated disparity map. The process consists of four rigorously defined and lightweight modules that can be iterated multiple times: a disparity cross check, bitwise fast voting, invalid disparity handling, and median filtering. We demonstrate that our iterative refinement has a large effect on the overall quality, resulting in smooth disparity maps with sharp object edges, especially around occluded areas. It can be applied to any stereo matching algorithm and tends to converge to a final solution. Finally, we perform a quantitative evaluation on various Middlebury datasets, showing an increase in quality of over several dB PSNR compared with their ground truth. Our whole disparity estimation algorithm supports efficient GPU implementation to facilitate scalability and real-time performance." "Capturing the complexity of integrated energy systems through model linking" "Matija Pavičević" "This research aims to develop a more efficient and effective way to plan and manage the integration of variable renewable energy sources (VRES) into the power grid. The approach involves developing a multi-sectoral unit commitment and energy dispatch (UCED) model, which optimally dispatches all cross-sectoral and power generation technologies and preallocates the state of charge of all storage technologies to ensure a reliable and cost-effective supply. A soft-linking framework that links the multi-sectoral UCED model with the Energy System Optimization Model (ESOM) is also developed, which can provide valuable information on the energy demand and supply, as well as the economic and environmental impacts of different energy scenarios. The research objectives of this study are: 1) to develop a multi-sectoral UCED model, 2) to investigate the impact of different model formulations on the speed and accuracy of the UCED model, 3) to develop and generalize a uni-directional and bi-directional soft-linking methodology capable of linking any ESOM model to the UCED model, 4) to apply the proposed uni-directional soft-linking methodology to a case study involving the JRC-EU-TIMES ESOM model and Dispa-SET UCED model, 5) to apply the proposed bi-directional soft-linking methodology to a case study involving the EnergyScope ESOM model and Dispa-SET UCED model, and 6) to find a trade-off between computation time and accuracy of the bi-directional soft-linking methodology. The main research challenge addressed by this study is the integration of high shares of VRES into the energy system, which is one of the main challenges of the energy transition. The study is important because by optimally managing the integration of these sources, the models can reduce greenhouse gas emissions in a cost-effective way. The sector-coupling, which is the integration of different energy sectors such as power, heating, gas, chemicals, and transportation, can also contribute to reaching carbon neutrality by increasing the flexibility and efficiency of the overall system. The primary advantage of utilizing a multi-sectoral UCED model is the enhanced precision and accuracy in assessing the operation of the energy system, specifically with regard to cross-sectoral technologies and the integration of high levels of VRES. A secondary advantage is the capacity to effectively evaluate the interactions and energy flows between the electricity and non-electricity sectors, through the inclusion of detailed representations of storage and demand-side management technologies. This enables a comprehensive understanding of the dynamic and complex relationships between these sectors. Furthermore, the multi-sectoral UCED model allows for the optimization of resource allocation by incorporating both volume-based and cost-based constraints. This enables a comprehensive evaluation of the trade-offs between resource availability and cost. The soft-linking framework has the ability to link the multi-sectoral UCED model with the ESOM model in a uni-directional or bi-directional way with feedback loops. By linking the two models, the UCED model can use the outputs from the ESOM model to assess the reliability and stability of the system and make more accurate predictions and decisions. The feedback loops between the two models can also enable the continuous optimization of the energy system, which can help to achieve carbon neutrality. The study also focuses on investigating the impact of different model formulations on the speed and accuracy of the UCED model, as well as the impact of various factors on the outcome of uni-directional and bi-directional soft-linking. The goal is to determine if the overall impact is significant enough to potentially overcome the issue of undersizing the system when using the ESOM as a stand-alone tool. The case studies involving the JRC-EU-TIMES and EnergyScope demonstrate the advantages of using the uni-directional and bi-directional soft-linking." "A Study of Partial Synchronization in Networks of Delay-Coupled Systems" "Libo Su" "This thesis studies partial synchronization in networks of delay-coupled systems. Partial synchronization refers to the phenomenon that coupled systems can be grouped into clusters such that synchrony is only observed within each cluster. The coupling considered here contains a time-delay. A pattern of partial synchronization can be characterized by a partial synchronization manifold which is a linear invariant subspace of the state space of the network dynamics, corresponding to partially synchronous motion. The existence and asymptotic stability of such a manifold are required to observe partial synchronization in simulations and experiments. The latter guarantees that the manifold attracts the neighboring solutions. Therefore, the partial synchronization problem can be tackled in two steps: 1) identify partial synchronization manifolds; 2) analyze the stability of the partial synchronization manifolds. This thesis aims to provide an efficient method to characterize all partial synchronization manifolds and tractable stability conditions of partial synchronization manifolds. In return, the method and stability conditions provide insights into the relation between the dynamics of individual systems, coupling parameters, time-delays, and partial synchronization.First, a method is proposed to identify all partial synchronization manifolds in delay-coupled networks, using the information on network structure and coupling type. The method relies on a recently proposed existence criterion for partial synchronization manifolds. The criterion is expressed as a row-sum check on the ordered adjacency matrix (which represents the network structure). In this thesis, the criterion is extended so that the method is also applicable to networks with different systems, and the computational efficiency is significantly improved for the case where only the transmitted signals in coupling contain a time-delay. In addition, the method allows to automatically decompose the network dynamics into synchronization error dynamics describing the deviations of the system states within the clusters, and dynamics on manifolds describing the network dynamics after synchronization. The method is implemented as a publicly available software tool.Second, the stability of the partial synchronization manifolds is analyzed by using the second method of Lyapunov. The stability analysis of partial synchronization manifolds is recast as the stability analysis of an equilibrium (or zero solution) of the synchronization error dynamics. It is shown that the linearized error system can be interpreted as a linear parameter-varying (LPV) time-delay system. By assessing the stability of this LPV time-delay system using a Lyapunov-Krasovskii functional (LKF), the sufficient conditions for the local stability of partial synchronization manifolds are derived and expressed in the form of linear matrix inequalities (LMIs), whose feasibility can be assessed by using standard numerical tools.Third, the partial synchronization in a practical sense is considered. From an application perspective, the conditions for partial synchronization are too stringent, as inevitable perturbations or uncertainties (considered as additive to the nominal setting) destroy the zero equilibrium of the error system. Consequently, exact partial synchronization, where the states of the systems within each cluster are perfectly synchronized, cannot be achieved. Instead, an approximate synchronization, practical partial synchronization, may be observed, where the states of the systems within each cluster converge to each other up to some tolerance, and this tolerance tends to zero if (the size of) the perturbations tend to zero. In this case, the error system in the analysis is viewed as a non-autonomous time-delay system with a bounded additive perturbation. By assessing the practical stability of this error system using LKFs, the conditions for practical partial synchronization is derived and again formulated in term of LMIs. Besides, an explicit relation between the size of perturbation and the bound of the synchronization error is provided. All these conditions (for both exact and practical synchronization) induce restrictions on the coupling strength and time-delay. They can be used to check whether for a given set of parameters partial synchronization occurs and shed a light on the parameter dependence.Fourth, experiments on a network of electronic neurons are performed to demonstrate the obtained theoretical results. Each electronic neuron is the implementation of a neuron model in a circuit board. A cross correlation-based notion of practical synchronization is introduced to quantify the synchronization between the neurons. The experiments confirm the theory well though the theoretical results tend to be conservative. The synchronization regions in network parameter space determined by the theoretical conditions are a qualitative match of those obtained via experiments, although the former are smaller than the latter due to the conservatism of the theory.Finally, a control design is presented to achieve vehicle platooning which is a real-life application of controlled network synchronization. The design, grounded in model predictive control (MPC), employs a cooperative adaptive cruise control (CACC) strategy to achieve vehicle platooning. The CACC strategy uses vehicle-to-vehicle communication to allow for inter-vehicle data exchange such that short inter-vehicle distances can be reached while guaranteeing safety. To improve the cohesion of the platoon when encountering disturbance, a bi-directional network topology is used. It is shown that compared to the commonly used unidirectional topology, the bi-directional topology allows the vehicles to maintain shorter distances." "Optical path characterization and optimization improves SWIR hyperspectral imaging of fruits" "Janos Keresztes" "Quality assessment and process monitoring are essential for today's fruit industry sector and the world's economy. From picking fruit in orchards, to transport and handling practices, to storage and packaging, each step will influence the quality of the end-product when presented to the consumers. The appearance, consisting of the shape, size, colour or absence of any damage are essential criteria relevant to consumers, and influence their will on buying. It is therefore essential to provide a fast, consistent quality assessment of each fruit to match the expectations of the market. To remain competitive to this demand at low cost, fast and efficiently, non-destructive automated quality sorting lines are needed. Among the different defects affecting fruit quality, bruises are one of the most problematic industrial losses. The detection of bruises in fruit such as apples during handling is therefore required. The browning process of bruises results in progressive apple tissue softening and colour changes. As this natural process takes time before it becomes visible, there is a gap of a few days between the mechanical damage causing the bruise and the consequent visible brown spots, which lower the price of apples. It is therefore important to detect bruises on each apple as early as possible after damaging to limit consequent economical losses. Light is the fastest known information carrier. In ambient conditions, it is also harmless for fruit or the surrounding workers, and a cheap technology. Among the different non-destructive and non-invasive techniques, the usage of light and the analysis of the information it can carry is therefore the most promising path. By shining light onto apples, and observing the absorbed, reflected and scattered light, bruises may be detected non-destructively at high speed. Among the most recent technologies reported, hyperspectral imaging (HSI), being the combination of the machine vision and spectroscopy fields, is showing promising paths. More particularly, the short-wave infra-red (SWIR) range has been demonstrated to promote successful detection of bruises in apples at early stages. However, there are still limiting factors when using SWIR HSI prior its success in industry. Among them, the most predominant are high noise levels arising from the detectors, non-uniform illumination, specular reflections and real-time HSI data handling. This research aims to tackle those problems be first describing and modelling the different components consisting of a SWIR HSI sorting system being the illumination and the imager, and further optimizes their configuration for better image quality. Those building blocks are further put together combined with improved data handling more robust and efficient usage of SWIR HSI in industry. This research is split into 10 chapters. Chapter 1 covers the current practices in image based fruit sorting, with a stronger emphasis on hyperspectral imaging. It further compares visible and near infra-red (Vis-NIR) to SWIR HSI and where are the additional challenges when using SWIR over Vis-NIR. The chapter then describes the relevance of apples in industry and why early bruise detection. The chapter ends with the outline of this thesis. Chapter 2 describes the state-of-the-art in light, its interaction with matter, with a focus on polarisation and vibrational spectroscopy. The browning bio-chemical process of apples is further described. The algorithms used to process light spectral information, also referred to multivariate data analysis which are applied within this research are then described. Chapter 3 is focused on characterising the noise and sensitivity of a SWIR hyperspectral imager, to quantify the signal to noise ratio (SNR). To quantify the pixel-to-pixel variation or the detector’s response, a radiometric calibration method is proposed which dynamically removes the detector noise. This approach removed 6% further noise compared to conventional sequential noise sampling methods. The average detector noise or dark current evolution through time is then shown, which was noted to vary non-linearly, with a sub-linear trend one hour after start-up to stabilize after 3 hours, with up to 12% of the imager’s dynamic range. Contrast of each spectral image is also described using a novel custom-made checkerboard calibration rig. The following showed a ratio per wavelength up to 15000 versus 1 raw values with 100-1700 nm. The checkerboard also enabled accurate spatial calibration using a thin lens model. Chapter 4 describes how to measure, model analytically and using non-sequential ray-tracing software the spectral and angular distribution of halogen tungsten (HT) spots, considered as the standard in SWIR HSI illumination. The far field angular distribution was modelled with a Gaussian distribution with an R² of 0.99, while the spectra using a Plank based 5th order polynomial with and R² of 0.98. The modelled spectra enabled to convert photometric measurements into radiometric units, and estimate the energy contribution of the spots in the SWIR spectral range, with up to 63% of the total spectral power. Further, near-field spot distribution is measured and modelled within the ray-tracing software, comparing irradiance distributions when using or not diffusers. It was shown that the irradiance patterns could be reproduced with a peak relative error of 12% when using diffusers, while up to 30% without. Conventional illumination distribution and light beam shaping are non-linear problems, which often are solved using iterative methods such as simplex or simulated annealing (SA) optimization algorithms, which can result in sub-optimal solutions or time consuming searches. Chapter 5 introduces novel constrained non-linear global optimization algorithms which can handle more efficiently such problems while simultaneously offering information on the sensitivity of a configuration near its optimum. A design is proposed using 4 HT spots placed around a flat target, using the source models from chapter 4. The two proposed optimization methods are referred to as Design of Computer Experiments with Design Augmentation (DACEDA) and Design of Computer Experiments with Simplex post-optimization (DACES). A 2 variables analytical version of the problem using isotropic source models enabled to compare DACES and DACEDA’s modelled design space with an overall average relative error of 2%, with a peak up to 10% at the corners of the design space. The SNR of ray-traced near-field sources modelled in chapter 4 is quantified using the Rose model, setting the stop criterion of the proposed optimization algorithms. The simulated irradiance distribution uniformity is then optimized for a 2 and 5 variables case studies with DACES, DACEDA, simplex and SA. In the 2D case, it was shown that DACES performed best after 30 simulations while in the 5D case, DACEDA performed best after 65 simulations. Both algorithms were further used in a case study with DACEDA for tolerance analysis, and DACES for optimization of a configuration for apples, which was used within the remaining of the thesis. Among the main challenges when using SWIR HSI for fruit quality inspection, are the glossy regions observed from their arbitrary deformed toroidal shape and waxy surface. Therefore, this research further aimed at reducing the influence of those specular reflections, both numerically and optically. In chapter 6, a first proposed approach is to use chemometrics tools combined with image analysis to reduce or remove those artefacts, using a multiclass classifier or a stepwise approach. The proposed method using iterative steps to remove progressively automatically unwanted regions resulted in 6% higher prediction accuracy than a multi-class partial least-squares discriminant analysis (PLS-DA) classifier. Appropriate wavelength selection using interval PLS-DA enabled to improve further by 4%. Furthermore, the stepwise algorithm enabled to detect for multiple cultivars up to 80% six hours after bruising. Chapter 7 uses the multi-class PLS-DA classifier from chapter 6 on a real-time case study of one cultivar, and compare pixel based calibration models to conventionally used region based ones. Pixel based models, encountering for variations described in chapter3, improved prediction accuracy at the pixel level up to 2%. With a cultivar based model built for 2 hours after damage, using area normalization as spectral pre-processing and image post-processing, a pixel-based prediction of accuracy of 95.6% was obtained, while up to 98% at the sample level. The following was demonstrated on a real-time SWIR HIS sorting system at a rate of 200 ms per apple at a scanning speed of 0.3 m/s. To further improve those results, chapter 8 aims at quantifying the degree of glossiness for apples as a function of the light geometrical path, also referred to as surface scatter properties or bi-directional scatter distribution function (BSDF). It was shown that apples have a Gaussian gloss trend around the specular angle, and are Lambertian outside the glossy region. Moreover, polarization properties of apples are then investigated in chapter 9, in the aim to remove optically gloss arising from apples using a cross-polarized imaging system. It was shown that gloss could be removed for multiple cultivars using cross-polarization, and that the resulting scattered reflected light was Lambertian, thus improving the image uniformity and bruise-sound contrast region. Finally, the combination of the conclusions drawn from the different chapters and future research perspectives are given in chapter 10. It can be concluded that near-field ray-traced sources with diffusers are the best choice for SWIR illumination, which can be optimized using DACES or DACEDA for improved uniformity. It was shown in this research that real-time SWIR HSI is possible, and using broad spectral has significant added value. It was shown that using the knowledge of polarization and surface scatter properties of apples, linear cross-polarized imaging configuration is the most efficient solution to remove gloss of fruit in images. Alternatives are also possible, by numerically removing gloss using area-normalization and PLS-DA, with image post-processing, which are cheaper, but more time consuming." "The role of school‐based relationships for school well‐being: How different are high‐ and average‐ability students?" "Tessa Weyns, Hilde Colpin, Karine Verschueren" "BACKGROUND: Relationships with peers and teachers are keys for children's psychosocial development. However, it is unclear whether this also applies for high-ability children. This study adds to the literature by longitudinally examining the role of interpersonal relationships in the educational context for high- and average-ability students. AIMS: In this study, we examined whether high- and average-ability students differ in terms of peer acceptance, teacher conflict, and school well-being. Further, we studied the potential bi-directional effects between school well-being, teacher conflict, and peer acceptance in late childhood and examined whether the relations between these concepts differ between high- and average-ability students. SAMPLE: The total sample consisted of 3,101 Belgian students (49.9% boys, Mage  = 9.76 years), with 348 high-ability and 2,753 average-ability students. METHODS: Cognitive ability was assessed in Grade 3 with two cognitive ability tests (for crystallized intelligence and fluid intelligence, respectively). The students were followed for three years (Grades 4, 5, and 6), and each year their current teacher filled out questionnaires regarding teacher-child conflict, peer acceptance, and school well-being. RESULTS: High-ability students showed lower teacher conflict, higher peer acceptance, and better school well-being than average-ability students. Cross-lagged analyses showed that peer acceptance consistently predicted school well-being over time, while Grade 4 school well-being impacted Grade 5 teacher conflict. Also, better peer acceptance in Grade 5 predicted lower teacher-student conflict in Grade 6. Multigroup analyses revealed similar associations for both groups. CONCLUSIONS: Our results show that high-ability students in general display more positive school-based social relationships and school well-being than average-ability students. We found no differences in the links between these variables for both groups (i.e., high-ability students and average-ability students)." "Neonatal EEG sleep stage classification based on deep learning and HMM" "Katrien Jansen, Maarten De Vos" "OBJECTIVE: Automatic sleep stage scoring is of great importance for investigating sleep architecture during infancy. In this work, we introduce a novel multichannel approach based on deep learning networks and hidden Markov models (HMM) to improve the accuracy of sleep stage classification in term neonates. APPROACH: The classification performance was evaluated on quiet sleep (QS) and active sleep (AS) stages, each with two sub-states, using multichannel EEG data recorded from sixteen neonates with postmenstrual age of 38-40 weeks. A comprehensive set of linear and nonlinear features were extracted from thirty-second EEG segments. The feature space dimensionality was then reduced by using an evolutionary feature selection method called MGCACO (Modified Graph Clustering Ant Colony Optimization) based on the relevance and redundancy analysis. A bi-directional long-short time memory (BiLSTM) network was trained for sleep stage classification. The number of channels was optimized using the sequential forward selection method to reduce the spatial space. Finally, an HMM-based postprocessing stage was used to reduce false positives by incorporating the knowledge of transition probabilities between stages into the classification process. The method performance was evaluated using the K-fold (KFCV) and leave-one-out cross-validation (LOOCV) strategies. MAIN RESULTS: Using six-bipolar channels, our method achieved a mean kappa and an overall accuracy of 0.71-0.76 and 78.9%-82.4% using the KFCV and LOOCV strategies, respectively. SIGNIFICANCE: The presented automatic sleep stage scoring method can be used to study the neurodevelopmental process and to diagnose brain abnormalities in term neonates." "Bio-ecological processes in schools: The role of teachers, peers, and genes in children's and adolescents' psychological and academic development" "Steven De Laet" "Throughout late childhood and adolescence, there is a general increase in adjustment problems, including an increase in psychosocial problems, such as externalizing behavior, and an increase in academic problems, such as declining levels of behavioral school engagement. The present doctoral dissertation offers a comprehensive framework for understanding the development of externalizing behavior and behavioral engagement, guided by the bio-ecological model of human development (Bronfenbrenner & Morris, 2006), which explains individual differences in development in terms of both environmental and biological influences and interactions between them. We focused on how the affective quality of teacher-student relationships, peer likeability and peer popularity impact children’s and adolescents’ behavioral school engagement and externalizing behavior. In addition, we revealed the bi-directional effects and transactional processes between teacher-student relationships and peer relationships in elementary school, and between teacher-student relationships and adolescent adjustment problems in secondary school. Based on the biological side of Bronfenbrenner’s model – which suggests that biologically-based factors may join with environmental factors to shape development – we investigated whether dopaminergic genes moderate the effect of teacher-student relationships on adolescents’ adjustment problems. Four empirical studies are included. The first two are based on a longitudinal project which includes 586 children from Grade 4, followed for three years (Sample 1). The second two are based on the longitudinal project Studying the Transition to Adolescence: Testing Genes, Interactions, and Environments (STRATEGIES) which includes 1116 adolescents from Grade 7 to 9, followed for three years (Sample 2). Several self-reported, parent-reported, and peer-reported measures were used to assess our research variables. To analyse the data, we used parallel latent growth curve analyses, cross-lagged analyses, multigroup analyses, and hierarchical regression analyses in the framework of structural equation modelling (SEM) using Mplus version 6 (Muthén & Muthén, 1998-2010). In the first study, we examined the reciprocal associations among teacher-student and peer relationships in late elementary school (Sample 1) and found that higher peer likeability predicted more teacher-student support, which in turn predicted higher peer likeability, beyond changes in children’s prosocial behavior. Higher peer popularity predicted more teacher-student conflict (driven by children’s aggressive behavior), which, in turn and in itself, predicted higher perceived popularity. The second study investigated whether teacher-student and peer relationships may buffer or exacerbate the normative declines in behavioral engagement in late elementary school (Sample 1)and showed that high initial and increasing levels of teacher-student support and high initial levels of peer likeability independently reduce the normative declines in children’s behavioral engagement. Children who were perceived as more popular in Grade 5 were less engaged in school in Grade 6. The third study aimed to test whether adolescents’ genotype for the dopamine transporter (DAT1) or receptor (DRD4) moderates the effect of teacher-student relationship quality on adolescents’ rule-breaking behavior and behavioral engagement (Sample 2). It seems that DAT1-10R homozygotes were more affected by the negative effect of lower teacher-student affiliation on behavioral engagement and that DRD4 non-long carriers were more affected by the adverse effect of teacher-student dissatisfaction on rule-breaking behavior. In the fourth study (Sample 2), we examined bidirectional relations between teacher-student relationship quality and adolescent rule-breaking behavior and behavioral engagement throughout secondary school. In addition, we investigated whether the effects of teacher-student relationship quality on subsequent adolescent behavior were moderated by a biologically informed polygenetic score (BIMPS) for dopamine signalling. We found that adolescents who displayed more rule-breaking behavior and less behavioral engagement experienced increased subsequent dissatisfaction with their teachers, which in turn further increased their adjustment problems. Also, adolescents with more adjustment problems experienced decreased subsequent affiliation with their teachers. The other way around, adolescents’ behavioral engagement also benefitted from positive relationships with teachers. In addition, adolescents who had a BIMPS score coding for moderate levels of dopamine signalling (instead of high or low signalling) were most affected in their behavioral engagement when they experienced dissatisfaction with their teachers." "Towards Robotic Architecture that Mediates Place: Prototyping Meaningful Human-Building Interactions through Autonomous Spatial Adaptation" "Alex Nguyen" "The discipline of adaptive architecture explores how the architectural space can dynamically adapt to the changing needs of occupants. Over five decades of research, it has proposed that such dynamic adaptations hold the potential to evoke 'poetic' experiences from occupants that might be as compelling as those of its 'static' counterpart. While it is still unclear how these experiences can unfold in a meaningful way, applicable theories from Environmental Psychology proposed that the meaningful human experience of a place depended on not only the spatial qualities of its physical manifestation, but also the situational qualities of its surrounding contexts and its subjective qualities as perceived by the occupants. Motivated by these insights, this dissertation investigated the research question: How can spatial adaptation, i.e. the act of physically adapting a spatial layout through the integration of robotic technology, dynamically satisfy the qualities of a place in a meaningful way? To adequately address its research question, this dissertation proposed that spatial adaptation should be designed from a phenomenological perspective that respects and exploits the unique qualities of how architectural space is holistically experienced. By applying evaluation methodologies from both Architectural Design and Human-Computer Interaction (HCI) research, the main contributions of this thesis are: 1) a series of working adaptive architecture prototypes; 2) a series of controlled and in-the-wild studies in varying real-world contexts; and 3) a collection of design considerations that inform the future research of adaptive architecture, robotic furniture, as well as other Human-Building Interaction (HBI) or Human-Robot Interaction (HRI). These contributions are manifested through seven peer-reviewed chapters clustered in three parts, among which: Part I. Exploration consists of two chapters that delved into the design space of spatial adaptation, offering methodological insights for the subsequent chapters. Chapter 2 investigated how lay-participants with no professional design background arranged interior layouts based on two locations of a 'static' partition in different realities. It discovered a range of spatial, situational, and subjective qualities that participants employed as design arguments for their preferred arrangements according to the locations of the 'static' partition. Chapter 3 compared how participants experienced two preliminary prototypes of spatial adaptation: a set of virtually simulated partitions versus a physical Wizard-of-Oz mobile robotic foldable partition. It captured a range of unique spatial qualities emerged from the dynamic movements of these spatial adaptation prototypes. Part II. Experimentation comprises three chapters that deployed spatial adaptation prototypes in real-world contexts through semi-controlled studies, providing insights into how people experienced spatial adaptation in the short term. Chapter 4 evaluated the semi-immersive, cross-reality simulation of an autonomous mobile partition adapting the homes of remote participants during the COVID-19 quarantine period. It provided preliminary insights into how and when spatial adaptation should occur based on a range of spatial and subjective qualities; and presented the first evidences indicating that participants were more accepting of unconventional spatial adaptations compared to identical 'static' layouts. Chapter 5 provided an intermediate reflection on previous chapters to discuss the research challenges associated with spatial adaptation, originated from the contrast viewpoints of ordinary users versus experts. Chapter 6 measured the impact of a Wizard-of-Oz mobile robotic partition in creating different 'places' for participants located in a breakout space of a shared office building. It proposed a preliminary sense of place framework that related these qualities to recommend how and when spatial adaptation should occur, proactively or reactively, to support one occupant. Part III. Implementation encompasses two chapters that described longer term, in-the-wild studies to validate previous findings while ensuring ecological validity in the research. Chapter 7 optimised the engagement of passers-by with a rhythm-making robotic facade, as it taught them how to engage through proactive feedforward rhythms, through which empowering them to improvise through reactive feedback rhythms. It demonstrated several behaviours of passers-by that can be tracked to understand and enhance their experience, which has the potentials for spatial adaptation to also apply. Chapter 8 evaluated the utilisation of a mobile robotic partition that semi-autonomously moved between different adaptations selected by participants to cope with spatial stressors in their real-life shared office, while taking into account the potential socio-spatial consequences to their colleagues. It identified four distinct adaptation strategies employed by participants, considering whether spatial adaptation should address a spatial stressor before or during its occurrence; and whether it should adapt the architectural zone of the participant in need or that of the spatial stressor. By relating the decision-making process of these adaptation strategies to the sense of place framework, this chapter clarified \textit{how} and \textit{when} spatial adaptation should occur for multiple occupants in a shared space. By synthesising insights from the aforementioned chapters, this dissertation offers a preliminary answer to the research question, addressing when and how spatial adaptation should occur, and highlighting its impact on occupants. From these insights, it proposes nine design considerations for future autonomous adaptive architecture, including: consider the perceived motivation and ability of the occupant to determine the appropriate moment to initiate a spatial adaptation, taking into account that the occupant can also employ subjective, situational, or synergistic adaptations to cope with the spatial stressor at hand; consider the architectural zone and the timing of the spatial adaptation, in relation to those of the occupant and the spatial stressor, as these factors result in different spatiotemporal adaptation strategies with perceivable distinct influences on the interdependent dimensions affecting the sense of place; and consider the characteristics of the spatial adaptation prototype at hand, which influenced each dimensions of the sense of place differently in a hierarchy, ultimately shaping its unique affordances perceived by co-located occupants sharing the same space. From these design considerations, this dissertation discussed the potential of adaptive architecture to leverage on its unique affordances to become a creative design dimension, potentially facilitating a truly bi-directional interaction loop between architectural space and its occupants. As the field of Evidence-Based Design proved that the built environment affected human health and well-being with longer-term implications that can even benefit the quality of life, this dissertation aims to inspire HBI as well as HRI research to ground new innovative interventions on the existing architectural knowledge to better benefit occupants in the longer term." "A Difference in What?" "Willy Ranson" "As Gregory Bateson noted [1], information can be described as 'a difference that makes a difference'. But a difference in what? Is it possible to communicate information between spatio-temporal locations as itself? It would seem not. We as humans communicate by speech, as a modulation of acoustic waves. We transmit images by modulating electro-magnetic waves. Information moves around in computers as a modulation of the clock frequency. Within a system's scalar level, information is exchanged as a modulation of properties of the level's population. Can information alone be communicated? No - as a 'difference', or as 'novelty', it is always superimposed on a structured 'carrier'. So let us look at the implications of this in different contexts.One of the present authors has published elsewhere details of the elastic similarity of zinc blende crystals of the group IV elements, the III-V compounds and the II-VI compounds [2]. Educational studies of crystallography would lead us to suppose that the directional elastic properties of macroscopic crystals should correspond exactly to those of their representative unit cells, but this is not strictly true. The correspondence is very nearly exact - to within less than ½ % for a comparison between the III-V compounds, for example - but there is a remaining degree of dependence on the atom types present. Elastic anisotropy in single crystals depends mainly on the basic order of the nominal crystal structure, but information about the atoms present is carried across from micro- to macro- scales, superimposed on the lattice structure: like a wireless transmission, it is transported 'on the back' of a carrier.It is interesting to note that in highly multi-scalar stems - i.e. uniquely biological ones - success in generating a 'higher' scale results in the imposition of downward 'slaving' on its progenital population, in turn strengthening the structural carrier which both permitted its emergence and transported the elements of its cross-scalar differentiation. The more levels that are generated, the more strongly lower levels will be slaved, only later resulting in their apparent structural stasis and 'simplicity'. It would be unwise, for example, to suppose that all electrons were identical in an early primeval post-bang soup - the simplicity of their contemporary low-quantum-number description is more likely the result of slaving from the multitude of higher scalar levels they now experience.We note, then, that this appears to be a general property of all information transport, whether in-scale or cross-scale: a structured carrier is always necessary to communicate 'a difference that makes a difference'. Bateson's 'difference' is a difference in a structured carrier.If we turn to bio-chemicals we find a similar situation. Based on the directional bonding of carbon, the smaller bio-molecules often exhibit nearly crystallographic atomic arrangements with only minor differentiating regions. A prime case is the lipid pdmpg, around 170 of whose some 200 molecular atoms are aligned in a tetrahedrally-regular 3D structure, leaving only a tail of 30 or so atoms to 'carry its message' to a higher scale. Ultimately, when we arrive at DNA, we find it to be a unified physically structured carrier - the double helix - which supports the dogmatically-supposedly 'sufficient' set of protein-representative genes (if we leave aside for the moment the last five-or-so years' revelations about conserved sequences and expression-control from within the 'junk' DNA!). It would not be stretching the point too much to describe DNA as a quasi-crystal, based on a quadruple distributed-unit-cellular basis of adenine, cytosine, guanine and thymine. In the massive proteins we find blossoming of this structural-carrier/differential-signal metaphor into the digital-analog bi-coding we associate with living tissues, where the combination of structural-bonding 'rigidity' and real-3D constraints produces the beauty of protein bending.But let us now take a step sideward into the properties of natural hierarchy. A radical difference between biological hierarchies and those we construct as engineers is that higher bio-levels operate faster than their constituent lower ones, and not more slowly as our own meager constructional experiences would lead us to expect. In addition, the natural world operates ecosystemically, and not mono-rationally as we ourselves aspire to do. We have published elsewhere extensive details of the bi-rational properties of natural hierarchies [e.g. 3, 4], where the correlated hierarchical scales we observe are interleaved with a second complementary correlated hierarchy which appears to consist of complexity ejected from the scales themselves. Within the observable Newtonian scales, information is naturally communicated as we described above - as quasi-disorder 'difference signals' superimposed on quasi-ordered 'carriers'. In the second, complementary hierarchy, however, we would expect to find information as quasi-ordered 'signals' superimposed on quasi-disordered 'carriers'. We now believe that, much as the observable scales can be reductively associated with 'matter', the complementary 'complex' scales may be associable with 'fields'.So what are the implications of all this to our understanding of our very selves, of our apparently dualistic nature of 'mind and matter'? First of all, we should remember that, much as biological evolution may 'wish' - whether by direction or stochastics - to implement a 'template' it finds all around in crystals and molecules, it is always dependent on what it already achieved - it cannot (easily!) 'start over again'. Consequently, although the relationship we find between unit cell and macroscopic III-V crystal may be elegant, evolution would be hard put to reproduce such a simplistic 'carrier-signal' relationship at the highest level of our being. And yet this seems to be the case.The two hemispheres of our brains appear to approximately correspond to the two interleaved 'structures' of a natural hierarchy: one is somewhat 'logical' (reductive towards localization), the other somewhat 'creative' (reductive towards delocalization). Most interestingly, although split-brain patients sometimes refer to a 'duality of consciousness', this is more generally a singular phenomenon, which is surprising for a brain which is divided into two parts! The hemispheres are connected by a massive 'information highway' - the corpus calossum. A number of researchers have suggested that the 40Hz EEG waves associated with consciousness couple the two hemispheres together through the corpus calossum - a reasonable conclusion. But is that all? Is it possible that these waves are primarily the carrier which supports the emergence of an integrated high level of awareness from brain-wide informational 'signal' fluctuations, thus delivering consciousness? If so, then life corresponds to a resonance between the quasi-ordered and quasi-disordered carriers of the two sub-hierarchical 'structures', and the evolution of high-level awareness in organisms would most likely follow that of the corpus calossum.[1] G. Bateson. Steps to an Ecology of Mind, pp. 457-459. U. C. Press, Chicago, 2000.[2] R. Cottam and G. A. Saunders. ""An Extension of Keyes' Correlation."" Physica Status Solidi (a) 33, 367-373, 1976.[3] R. Cottam, W. Ranson and R. Vounckx. ""Autocreative Hierarchy I: Structure - Ecosystemic Dependence and Autonomy."" SEED Journal 4, 24-41, 2003.[4] R. Cottam, W. Ranson and R. Vounckx. ""Autocreative Hierarchy II: Dynamics - Self-organization, Emergence and Level-changing."" In: H. Hexmoor (Ed.), International Conference on Integration of Knowledge Intensive Multi-Agent Systems, pp. 766-773. IEEE: Piscataway, NJ," "Extensions to Type Classes and Pattern Match Checking" "Georgios Karachalias" "Static typing is one of the most prominent techniques in the design of programming languages for making software more safe and more reusable. Furthermore, it provides opportunities for compilers to automatically generate boilerplate code, using mathematical foundations. In this thesis, we extend upon the design of Haskell, a general-purpose functional programming language with strong static typing, to offer more opportunities for reasoning, abstraction, and static code generation. More specifically, we improve upon two features: pattern matching and type classes. Pattern matching denotes the act of performing case-analysis on the shape of a value, thus enabling a program to behave differently, depending on input information. With the advent of extensions such as Generalized Algebraic Data Types (GADTs), pattern guards, and view patterns, the task of reasoning about pattern matching has become much more complex. Though existing approaches can deal with some of these features, no existing algorithm can accurately reason about pattern matching in the presence of all of them, thus hindering the ability of the compiler to guide the development process. The first part of this thesis presents a short, easy-to-understand, and modular algorithm which can reason about lazy pattern matching with GADTs and guards. We have implemented our algorithm in the Glasgow Haskell Compiler. The second part of this thesis extends upon the design of type classes, one of the most distinctive features of Haskell for function overloading and type-level programming. We develop three independent extensions that lift the expressive power of type classes from simple Horn clauses to a significant fragment of first-order logic, thus offering more possibilities for expressive type-level programming and automated code generation. The first feature, Quantified Class Constraints, lifts the language of constraints from simple Horn clauses to Harrop formulae. It significantly increases the modelling power of type classes, while at the same time it enables a terminating type class resolution for a larger class of applications. The second feature, Functional Dependencies, extends type classes with implicit type-level functions. Though functional dependencies have been implemented in Haskell compilers for almost two decades, several aspects of their semantics have been ill-understood. Thus, existing implementations of functional dependencies significantly deviate from their specification. We present a novel formalization of functional dependencies that addresses all such problems, and give the first type inference algorithm for functional dependencies that successfully elaborates the feature into a statically-typed intermediate language. The third feature, Bidirectional Instances, allows for the interpretation of class instances bidirectionally, thus indirectly adding the biconditional connective in the language of constraints. This extension significantly improves the interaction of GADTs and type classes, since it allows functions with qualified types to perform structural induction over indexed types. Moreover, under this interpretation class-based extensions such as functional dependencies and associated types become much more expressive. In summary, this thesis extends upon existing and develops new type-level features, promoting the usage of rich types that can capture and statically enforce program properties."