Title Promoter Affiliations Abstract "End-to-end Training of Deep Neural Networks for Monocular Autonomous Navigation - A Feasibility Study" "Tinne Tuytelaars" "ESAT - PSI, Processing Speech and Images" "This PhD investigates the feasibility of applying Deep Learning methods to monocular autonomous navigation. Deep Learning has shown impressive results over the last seven years in computer vision, moving from hand crafted algorithms to end-to-end learned models. Whether the same idea of moving from hand crafted algorithms to end-to-end trained deep neural networks is also applicable to monocular navigation is still an open question.This dissertation answers this question by first building an integrated software framework for imitation learning in simulation where extra sensors can be deployed for supervision. In a second step, the design decisions related to applying Deep Learning to monocular autonomous navigation is disentangled over three sets of design decisions: neural network architecture, training algorithm and domain generalization solutions. The impact of each design decision is evaluated on performance, training stability and sample efficiency on a corridor following task in simulation.The first set covers the neural network architecture where networks of varying depths, widths, input and output domains are compared. In order to train recurrent neural networks, a more robust training procedure is proposed, called Windowwise-Backpropagation Through Time.The second set of design decisions discusses the imitation learning strategy by:- changing the data collection procedure with different policies,- introducing a new online continual learning method for training without requiring stored datasets,- introducing a new sample efficient reinforcement learning method for self-supervised collision avoidance without requiring actual collisions.The third set of design decisions summarizes and evaluates four solutions for dealing with the domain shift. This domain shift occurs when a neural network trained for one environment has to perform in a different environment. In this case, we look at the shift between the simulated environment and the real world. The first solution uses domain invariant intermediate representations, in this case depth estimations. The second solution covers domain randomization. By adding a large variety in training simulated environments, the model successfully generalizes to a new very different environment such as the real world. The third solution implements auxiliary tasks, such as depth prediction, to counter environment specific overfitting. The fourth solution transfers the style of the simulated data to look similar to real-world data.Comparing learning based methods in robotics is difficult due to a lack of benchmarks. This thesis provides such comparison by investigating the impact of many design decisions on performance, training stability and sample efficiency. Despite the preliminary nature of the experimental setup, namely corridor following, the opinions and conclusions regarding architecture, data sampling and generalization is crucial to take the step to more complex autonomous navigation tasks. " "IWT-TETRA-project 090144 deLearyous: ""Training of interpersonal communication by natural language interaction with autonomous virtual characters""" "Koen Eneman" "Electrical Engineering Technology (ESAT), Group T Leuven Campus" "The aim of the deLearyous research project is to develop an interactive serious 3D game for the training of interpersonal communication skills within a professional context, eg. in an employer-employee or customer-employee relationship. The game consists in letting the trainee interact with autonomous virtual characters, which respond to the input of the player in a realistic and expressive manner. This allows the trainee to practice different behaviors and roles at his/her own pace, in a safe and virtual environment." "The influence of the multi-level environment and globalization on (autonomous) public sector organizations." "Koen Verhoest" "Public Administration & Management" "CentraI is the question how public sector organisations are steered and coordinated in a multi-Ievel environment (and how the control by different governmental levels interact with each other). Additionally, affiliated research issues are defined which focus on related issues of the influence of global crises, risks and reform ideas on governance, on multi-Ievel collaboration and regulation,and on national public administrations. The research plan aims at designing and testing refined explanatory modeis, drawing from different theoretical perspectives, by combining quantitative and qualitative methodological approaches." "The offender’s experience of punitiveness : a descriptive phenomenological study" "Ivo Aertsen" "Leuven Institute of Criminology" "Imposing a penal sanction generally implies the intentional infliction of suffering or pain on a person convicted of committing criminal offenses. On the one hand, this added suffering is being legitimized by penal goals, whether or not they are being made explicit, on the other hand this practice is being legitimized by respecting the principles of fair sentencing, such as proportionality and equality. When determining the punishment and the amount of punishment, one usually supposes that the suffering they intentionally inflict corresponds to the suffering that is effectively experienced by the punished. However, various studies show that this line of reasoning does not hold true and that the personal and social reality of being punished often differs from the objective intention (for example Crouch, 1993; Durnescu, 2011; Hayes, 2015, 2018b, 2018a; Kolber, 2009, 2012; van Ginneken & Hayes, 2017).While most research concerning the subjective experience of punishment focuses on charting the experienced suffering (the 'pains') of punished persons or the perceived severity of punishment, this study tries to map the perceived punitive character, more particular those elements that are being considered as the punitive components of the punishment by the one who is being punished. Using the descriptive phenomenological method developed by Dr. Amedeo Giorgi (1997, 2009, 2012), information is collected in a very inductive (intuitive) way about the experienced punitive character of various punishments, including imprisonment, (autonomous) electronic supervision and the community service order. The interviews with 14 (ex-) punished respondents are carefully analyzed according to the descriptive phenomenological method.The results show a great amount of variety when it comes to the experienced punitive character. Nonetheless, some common elements amongst all or most respondents were found, which led to the identification of three ‘structures’ that try to capture the essence of ‘being punished’ (without aiming at generalization). For the respondents, regardless of the sentence they were undergoing, 'being punished' means first of all feeling limited in freedom, which could be an experienced loss of time, money, physical freedom, freedom of action, autonomy and/or personal agency. The nature and the intensity of the experienced restriction of freedom depended on the sentence and personal and contextual characteristics of the offender. Secondly, ‘being punished’ meant for most of the respondents experiencing a negative emotion as a result of the criminal intervention (which one cannot control). Some of the respondents had a punishment experience as a result of feeling subjected to power or control, of being scared or insecure concerning their release from prison, of having the idea that their relatives were being punished as well, or as a result of experiencing stigma. Finally, respondents felt punished when they were experiencing a sense of injustice regarding the execution of the sentence and / or the consequences of the sentence. Most respondents indicated that they considered their sentence as a ‘logical’ and ‘deserved’ consequence of their actions. Despite their acceptance of guilt, of their sentence, they also experienced a certain injustice which, for them, increased the severity of their punishment or was experienced as an (additional) punitive element. They experienced an injustice concerning the criminal proceedings, certain final decisions, the sentencing process, the execution of the sentence or certain unintentional consequences of their sentence. In addition to the identification of these three ‘structures’ that formed the essence of ‘being punished’ for the participating respondents, the study provided some insights that could at least give rise to further investigation. It turned out that whenever the respondents felt punished, they experienced a certain suffering, a certain pain, as a result of the criminal intervention. In other words, the experienced suffering seems to be a condition for being able to speak of a punitive experience, but not every experienced suffering constitutes a punitive experience. It is therefore possible to reflect on the extent to which the experienced suffering should be taken into account when estimating the severity of the punishment. In addition, several respondents spontaneously expressed their opinion on the severity of the punishment they experienced. The degree of freedom restriction experienced by them was not always consistent with the ‘punishment hierarchy’ used in the penal code. For example, imprisonment was not always perceived as the most severe punishment and, for some respondents, preference was given to the prison sanction over electronic monitoring, an autonomous community service (‘work penalty’) or a fine. Electronic monitoring was perceived by some respondents as the most severe punishment, as they felt more restricted in their freedom when undergoing this punishment than when serving a prison sentence. This finding is in line with earlier research findings in this regard (including Vanhaelemeesch, 2015). In addition, some respondents found the work penalty to be experienced as a ‘punishment’, but to provide a rather positive experience, in the sense that this punishment appears to respect or even promote the autonomy or agency of the punished.Finally, based on the research results, we can state that many respondents feel punished as a result of an experienced suffering that falls outside the intention of the criminal justice system. Ignoring this ‘unintentional’ suffering can be detrimental to a just and fair criminal justice system. In order to limit as much as possible the 'harmful' suffering, in particular that experienced (unintentional) suffering that is counterproductive for good reintegration and rehabilitation (De Vos & Gilbert, 2017), it might be an option to expand the ‘damage limitation principle’, so that the principle focuses not only on avoiding 'detention damage' during the execution of a prison sentence, but more generally on avoiding 'punishment damage', both in the sentencing and execution phase."