This thesis principally addresses some problems in genetic programming (GP) and grammar-guided genetic programming (GGGP) arising from the lack of operators able to make small and bounded changes on both genotype and phenotype space. It proposes a new and flexible representation for genetic programming, using a state-of-the-art formalism from natural language processing, Tree Adjoining Grammars (TAGs). It demonstrates that the new TAG-based representation possesses two important properties: non-fixed arity and locality. The former facilitates the design of new operators, including some which are bio-inspired, and others able to make small and bounded changes. The latter ensures that bounded changes in genotype space are reflected in bounded changes in phenotype space. With these two properties, the thesis shows how some well-known difficulties in standard GP and GGGP tree-based representations can be solved in the new representation. These difficulties have been previously attributed to the treebased nature of the representations; since TAG representation is also tree-based, it has enabled a more precise delineation of the causes of the difficulties. Building on the new representation, a new grammar guided GP system known as TAG3P has been developed, and shown to be competitive with other GP and GGGP systems. A new schema theorem, explaining the behaviour of TAG3P on syntactically constrained domains, is derived. Finally, the thesis proposes a new method for understanding performance differences between GP representations requiring different ways to bound the search space, eliminating the effects of the bounds through multi-objective approaches.

Realistic visual and audio rendering still remains a technical challenge. Indeed, typical computers do not cope with the increasing complexity of today's virtual environments, both for audio and visuals, and the graphic design of such scenes require talented artists. In the first part of this thesis, we focus on audiovisual rendering algorithms for complex virtual environments which we improve using human perception of combined audio and visual cues. In particular, we developed a full perceptual audiovisual rendering engine integrating an efficient impact sounds rendering improved by using our perception of audiovisual simultaneity, a way to cluster sound sources using human's spatial tolerance between a sound and its visual representation, and a combined level of detail mechanism for both audio and visuals varying the impact sounds quality and the visually rendered material quality of the objects. All our crossmodal effects were supported by the prior work in neuroscience and demonstrated using our own experiments in virtual environments. In a second part, we use information present in photographs in order to guide a visual rendering. We thus provide two different tools to assist “casual artists” such as gamers, or engineers. The first extracts the visual hair appearance from a photograph thus allowing the rapid customization of avatars in virtual environments. The second allows for a fast previewing of 3D scenes reproducing the appearance of an input photograph following a user's 3D sketch. We thus propose a first step toward crossmodal audiovisual rendering algorithms and develop practical tools for non expert users to create virtual worlds using photograph's appearance.

Pour résoudre les problèmes lies au trafic important dans les métropoles, de nouvelles alternatives, appelées "Urban Transportation Systems", voient le jour. Celles-ci sont basées sur des véhicules électriques en libre accès. Une fonctionnalité nécessaire de ces systèmes est leur capacité a' se déplacer en convoi. Dans ce manuscrit de thèse, le travail présenté traite de la commande d'un convoi de véhicules urbains. Dans ce mémoire, une modélisation cinématique de type tricycle est adoptée pour représenter le robot mobile. La transformation en système chaîné de ce modèle permet de découpler les commandes latérale et longitudinale. La commande latérale basée sur des techniques de linéarisation exacte, permet d'assurer un suivi de trajectoire avec une tr'es haute précision (puisque les non-linearites du modèle ne sont pas approximées mais au contraire, explicitement prises en compte), indépendamment de la vitesse du robot. La loi de commande longitudinale, basée également sur des techniques de linéarisation exacte, suit une stratégie de commande globale : l''état du convoi est pris en compte, et non pas seulement l'état du véhicule situe immédiatement devant. En fait, le comportement du convoi est étudié à travers les états du leader et du véhicule précédent. L'écart inter-vehicules a réguler est choisi égal à la distance curviligne parcourue le long d'un chemin de référence, ce qui permet d'assurer un comportement cohérent du convoi, y compris lors du suivi de trajectoires a' forte courbure. Afin de prodiguer un confort suffisant aux passagers et d'assurer un fonctionnement en toute sécurité, des fonctions de supervision sont greffées aux lois de commande nominales, d'écrites ci-dessus. Ce module de supervision, nomme "Monitoring", gère également l'insertion ou la désinsertion de véhicules au sein du convoi. Des expérimentations, portées sur des voitures urbaines réelles, nommées Cycabs, permettent de valider les performances de la commande. Pour cela, les véhicules sont équipes d'une communication sans fil de type WiFi. Deux modalités capteurs ont 'et'e envisagées pour localiser les véhicules : principalement un capteur RTK-GPS (Real Time Kinematic Global Positioning System), mais également un système de vision monoculaire.

<p>In this project several tests were performed to evaluate the performance of an On-Board Imager® (OBI) mounted on a clinical linear accelerator. The measurements were divided into three parts; geometric accuracy, image registration and couch shift accuracy, and image quality. A cube phantom containing a radiation opaque marker was used to study the agreement with treatment isocenter for both kV-images and cone-beam CT (CBCT) images. The long term stability was investigated by acquiring frontal and lateral kV images twice a week over a 3 month period. Stability in vertical and longitudinal robotic arm motion as well as the stability of the center-of-rotation was evaluated. Further, the agreement of kV image and CBCT center with MV image center was examined.</p><p>A marker seed phantom was used to evaluate and compare the three applications in image registration; 2D/2D, 2D/3D and 3D/3D. Image registration using kV-kV image sets were compared with MV MV and MV-kV image sets. Further, the accuracy in 2D/2D matches with images acquired at non-orthogonal gantry angles was evaluated. The image quality in CBCT images was evaluated using a Catphan® phantom. Hounsfield unit (HU) uniformity and linearity was compared with planning CT. HU accuracy is crucial for dose verification using CBCT data.</p><p>The geometric measurements showed good long term stability and accurate position reproducibility after robotic arm motions. A systematic error of about 1 mm in lateral direction of the kV-image center was detected. A small difference between kV and CBCT center was observed and related to a lateral kV detector offset. The vector disagreement between kV- and MV-image centers was  2 mm at some gantry angles. Image registration with the different match applications worked sufficiently. 2D/3D match was seen to correct more accurately than 2D/2D match for large translational and rotational shifts. CBCT images acquired with full-fan mode showed good HU uniformity but half fan images were less uniform. In the soft tissue region the HU agreement with planning CT was reasonable while a larger disagreement was observed at higher densities. This work shows that the OBI is robust and stable in its performance. With regular QC and calibrations the geometric precision of the OBI can be maintained within 1 mm of treatment isocenter.</p>

Ce document constitue une synth`ese de travaux de recherche en vue de l'obten- tion du diplˆome d'habilitation `a diriger les recherches. A la suite ce cette in- troduction r ́edig ́ee en franc ̧ais, le reste de ce document sera en anglais. Je suis actuellement charg ́e de recherches INRIA au centre de Rennes Bretagne Atlantique. J'ai rejoint en Septembre 2001 l' ́equipe Vista dirig ́ee par Patrick Bouthemy, puis l' ́equipe Visages dirig ́ee par Christian Barillot en Janvier 2004. Depuis Janvier 2010, je travaille dans l' ́equipe-projet Serpico dirig ́ee par Charles Kervrann dont l'objet est l'imagerie et la mod ́elisation de la dynamique intra- cellulaire. Parmi mes activit ́es pass ́ees, ce document va se concentrer uniquement sur les activit ́es portant sur la neurochirurgie guid ́ee par l'image. En parti- culier, les travaux effectu ́es sur le recalage non-rigide ne seront pas pr ́esent ́es ici. Concernant le recalage, ces travaux ont commenc ́e pendant ma th`ese avec le d ́eveloppement d'une m ́ethode de recalage 3D bas ́e sur le flot optique [72], l'incorporation de contraintes locales dans ce processus de recalage [74] et la validation de m ́ethodes de recalage inter-sujets [71]. J'ai poursuivi ces travaux apr`es mon recrutement avec Anne Cuzol et Etienne M ́emin sur la mod ́elisation fluide du recalage [44], avec Nicolas Courty sur l'acc ́el ́eration temps-r ́eel de m ́ethode de recalage [42], et sur l' ́evaluation des m ́ethodes de recalage dans deux contextes : celui de l'implantation d' ́electrodes profondes [29] et le re- calage inter-sujets [92]. L'utilisation de syst`emes dits de neuronavigation est maintenant courante dans les services de neurochirurgie. Les b ́en ́efices, attendus ou report ́es dans la litt ́erature, sont une r ́eduction de la mortalit ́e et de la morbidit ́e, une am ́elio- ration de la pr ́ecision, une r ́eduction de la dur ́ee d'intervention, des couˆts d'hospitalisation. Tous ces b ́en ́efices ne sont pas `a l'heure actuelle d ́emontr ́es `a ma connaissance, mais cette question d ́epasse largement le cadre de ce doc- ument. Ces syst`emes de neuronavigation permettent l'utilisation du planning chirurgical pendant l'intervention, dans la mesure ou` le patient est mis en cor- respondance g ́eom ́etrique avec les images pr ́eop ́eratoires `a partir desquelles est pr ́epar ́ee l'intervention. Ces informations multimodales sont maintenant couramment utilis ́ees, com- prenant des informations anatomiques, vasculaires, fonctionnelles. La fusion de ces informations permet de pr ́eparer le geste chirurgical : ou` est la cible, quelle est la voie d'abord, quelles zones ́eviter. Ces informations peuvent main- tenant ˆetre utilis ́ees en salle d'op ́eration et visualis ́ees dans les oculaires du mi- croscope chirurgical grˆace au syst`eme de neuronavigation. Malheureusement, cela suppose qu'il existe une transformation rigide entre le patient et les im- ages pr ́eop ́eratoires. Alors que cela peut ˆetre consid ́er ́e comme exact avant l'intervention, cette hypoth`ese tombe rapidement sous l'effet de la d ́eformation des tissus mous. Ces d ́eformations, qui doivent ˆetre consid ́er ́ees comme un ph ́enom`ene spatio-temporel, interviennent sous l'effet de plusieurs facteurs, dont la gravit ́e, la perte de liquide c ́ephalo-rachidien, l'administration de pro- duits anesth ́esiants ou diur ́etiques, etc. Ces d ́eformations sont tr`es difficiles `a mod ́eliser et pr ́edire. De plus, il s'agit d'un ph ́enom`ene spatio-temporel, dont l'amplitude peut varier consid ́era- blement en fonction de plusieurs facteurs. Pour corriger ces d ́eformations, l'imagerie intra-op ́eratoire apparait comme la seule piste possible.

Recent progress in the overlay and registration of digital information on the users workspace in a spatially meaningful way has allowed mixed reality (MR) to become a more effective operational medium. In the area of medical surgery, surgeons are conveyed with information such as the incisions location, regions to be avoided, diseased tissues, etc, while staying in and keeping their original working environment. The main objective of this Thesis is identifying theoretical and practical basis for how mixed reality interfaces might provide support and augmentation maximizing the continuity of interaction. We start proposing a set of design principles organized in a design space which allows to identify continuity interaction properties at an early stage of the development system. Once the abstract design possibilities have been identified and a concrete design decision has been taken, an implementational strategy can be developed. Two approaches were investigated: markerless and marker-based. The last one is used to provide surgeons with guidance on an osteotomy task in the maxillo-facial surgery. The evaluation process applies usability tests with users to validate the augmented guidance in different scenarios and to study the influence of different design variables in the final user interaction. As a result we have found a model to describe the contribution factors of each variable for the continuity of the user interaction. We suggest that this methodology can be applied mainly to those applications in which smooth connections and interactions, with virtual and real environments, are critical for the system; i.e. surgery, drivers applications or pilot simulations.

Automated Guided Vehicle (AGV) Container-Job deployment is essentially a vehicle-dispatching problem. In this problem, the impact of vehicle dispatching polices on the ship makespan for discharging and/or loading operations is analyzed. In particular, given a storage location for each container to be discharged from the ship and given the current location of each container to be loaded onto the ship, the problem is to propose an efficient deployment scheme to dispatch vehicles to containers so as to minimize the makespan of the ship so as to increase the throughput. The makespan of the ship refers to the time a ship spends at the port for loading and unloading operations. In this paper, we will compare the performance of current deployment scheme used with the new proposed deployment scheme, both with deadlock prediction & avoidance algorithm done in previous study [1]. The prediction & avoidance algorithm predicts and avoids cyclic deadlock. The current deployment scheme, namely pmds makes use of a greedy heuristics which dispatches the available vehicle that will reach the quay with the minimum amount of time the vehicle has to spend waiting for the crane to discharge/load the container from/onto the ship. The new deployment scheme, namely mcf aims to formulate the problem as a minimum cost flow problem, which will then be solved by network simplex code. The two simulation models are implemented using discrete-event simulation software, AutoMod, and the performances of both deployment schemes are analyzed. The simulation results show that the new deployment scheme will result in a higher throughput and lower ship makespan than the current deployment scheme. / Singapore-MIT Alliance (SMA)

Recent progress in the overlay and registration of digital information on the users workspace in a spatially meaningful way has allowed mixed reality (MR) to become a more effective operational medium. In the area of medical surgery, surgeons are conveyed with information such as the incisions location, regions to be avoided, diseased tissues, etc, while staying in and keeping their original working environment. The main objective of this Thesis is identifying theoretical and practical basis for how mixed reality interfaces might provide support and augmentation maximizing the continuity of interaction. We start proposing a set of design principles organized in a design space which allows to identify continuity interaction properties at an early stage of the development system. Once the abstract design possibilities have been identified and a concrete design decision has been taken, an implementational strategy can be developed. Two approaches were investigated: markerless and marker-based. The last one is used to provide surgeons with guidance on an osteotomy task in the maxillo-facial surgery. The evaluation process applies usability tests with users to validate the augmented guidance in different scenarios and to study the influence of different design variables in the final user interaction. As a result we have found a model to describe the contribution factors of each variable for the continuity of the user interaction. We suggest that this methodology can be applied mainly to those applications in which smooth connections and interactions, with virtual and real environments, are critical for the system; i.e. surgery, drivers applications or pilot simulations.

In this project several tests were performed to evaluate the performance of an On-Board Imager® (OBI) mounted on a clinical linear accelerator. The measurements were divided into three parts; geometric accuracy, image registration and couch shift accuracy, and image quality. A cube phantom containing a radiation opaque marker was used to study the agreement with treatment isocenter for both kV-images and cone-beam CT (CBCT) images. The long term stability was investigated by acquiring frontal and lateral kV images twice a week over a 3 month period. Stability in vertical and longitudinal robotic arm motion as well as the stability of the center-of-rotation was evaluated. Further, the agreement of kV image and CBCT center with MV image center was examined. A marker seed phantom was used to evaluate and compare the three applications in image registration; 2D/2D, 2D/3D and 3D/3D. Image registration using kV-kV image sets were compared with MV MV and MV-kV image sets. Further, the accuracy in 2D/2D matches with images acquired at non-orthogonal gantry angles was evaluated. The image quality in CBCT images was evaluated using a Catphan® phantom. Hounsfield unit (HU) uniformity and linearity was compared with planning CT. HU accuracy is crucial for dose verification using CBCT data. The geometric measurements showed good long term stability and accurate position reproducibility after robotic arm motions. A systematic error of about 1 mm in lateral direction of the kV-image center was detected. A small difference between kV and CBCT center was observed and related to a lateral kV detector offset. The vector disagreement between kV- and MV-image centers was  2 mm at some gantry angles. Image registration with the different match applications worked sufficiently. 2D/3D match was seen to correct more accurately than 2D/2D match for large translational and rotational shifts. CBCT images acquired with full-fan mode showed good HU uniformity but half fan images were less uniform. In the soft tissue region the HU agreement with planning CT was reasonable while a larger disagreement was observed at higher densities. This work shows that the OBI is robust and stable in its performance. With regular QC and calibrations the geometric precision of the OBI can be maintained within 1 mm of treatment isocenter.

<p>Ferroelectric thin films (FETFs) offer a promising new platform for advancing liquid-phase interfacial sensing devices. FETFs are capable of expressing surface charge densities that are an order of magnitude higher than those of traditional charged surfaces in liquid environments (e.g., common oxides, self-assembled monolayers, or electrets). Furthermore, the switchable polarization state of FETFs enables patterning of charge-heterogeneous surfaces whose charge patterns persist over a range of environmental conditions. Integration of FETFs into liquid-phase interfacial sensing devices, however, requires the fabrication of films with nanometer-scale surface roughness, high remnant polarization values, and interfacial stability during prolonged exposure. The objectives of my research were to i) fabricate ferroelectric ultra-smooth lead zirconium titanate (US-PZT) thin films with nanometer-scale surface roughness, ii) establish the interfacial stability of these films after prolonged exposure to aqueous environments, iii) measure the interfacial forces as a function of film polarization and ionic strength, iv) calculate the surface potential of the US-PZT surface using electric double layer (EDL) theory, and v) demonstrate the guided deposition of charged colloidal particles onto locally polarized US-PZT thin films from solution. </p><p>I demonstrate the use of ferroelectric US-PZT thin films to manipulate EDL interaction forces in aqueous environments. My work conclusively shows that the polarization state of US-PZT controls EDL formation and can be used to induce the guided deposition of charged colloidal particles in solution. </p><p>I present a robust fabrication scheme for making ferroelectric US-PZT thin films from a sol-gel precursor. By optimizing critical thermal processing steps I am able to minimize the in-plane stress of the film and reliably produce US-PZT thin films on the wafer-scale with mean surface roughness values of only 2.4 nm over a 25 &#956;m2 area. I then establish US-PZT film stability in water by measuring changes in film topography, crystallinity, surface chemistry, and electrical properties as a function of exposure duration. My results show that fabrication of crack-free US-PZT thin film is critical for long-term film fidelity in aqueous environments. Furthermore, I found no change in film topography or bulk composition with increasing exposure duration. Prolonged exposure to aqueous environments, however, gradually oxidizes the surface of the US-PZT wich results in a decrease in film resistivity and polarization saturation. Next, I used colloidal probe force microscopy (CPFM) to measure the EDL interaction force as a function of separation distance between polarized US-PZT thin films and a clean borosilicate probe. CPFM measurements were performed on oppositely polarized US-PZT thin films, which expressed either a positive or negative surface charge, and over a range of ionic strengths. The inner-Helmholtz plane (IHP) potential of the US-PZT was determined by fitting the CPFM force-separation data to number of EDL models, including; an analytical EDL model using a constant potential boundary condition with a Stern layer, a charge regulation EDL model, and a numerical EDL model using the non-linear Poisson-Boltzmann equation. Each model provides good agreement with the experimentally measured and predict high IHP surface potential for the polarized US-PZT thin films in solution. Finally, I demonstrate the use of polarized US-PZT to induce the guided deposition of positively or negatively charged colloidal particles from aqueous environments. I explore the effects of ionic strength, particle size, surface roughness, and pH on particle deposition. </p><p>Overall, this work demonstrates, for the first time, that FETFs can be used as a platform to manipulate colloidal particles in aqueous environments. The experimental results demonstrate that the surface charge of the FETF is reduced by charge shielding and perform similarly to traditional, charged surfaces in aqueous environments.</p> / Dissertation