Context. There have been a lot of progress made in the field of pathfinding. One of the most used algorithms is A*, which over the years has had a lot of variations. There have been a number of papers written about the variations of A* and in what way they specifically improve A*. However, few papers have been written comparing A* with several different variations of A*. Objectives. The objectives of this thesis is to find how Dijkstra's algorithm, IDA*, Theta* and HPA* compare against A* based on the variables computation time, number of opened nodes, path length as well as number of path nodes. Methods. To find the answers to the question in Objectives, an experiment was set up where all the algorithms were implemented and tested over a number of maps with varying attributes. Results. The experimental data is compiled in a table showing the result of the tested algorithms for computation time, number of opened nodes, path length and number of path nodes over a number of different maps as well as the average performance over all maps. Conclusions. A* is shown to perform well overall, with Dijkstra's algorithm trailing shortly behind in computation time and expanded nodes. Theta* finds the best path, with overall good computation time marred by a few spikes on large, open maps. HPA* performs poorly overall when fully computed, but has by far the best computation time and node expansion when partially pre-computed. IDA* finds the same paths as A* and Dijkstra's algorithm but has a notably worse computation time than the other algorithms and should generally be avoided on octile grid maps.

La durabilité chimique des alliages métalliques résulte notamment de la nature des défauts ponctuels assurant le transport au travers du film d’oxydation formé en surface. L'élaboration de couches d'oxyde modèles par oxydation thermique en pression contrôlée et ALD (Atomic Layer Deposition) et l'étude de leurs propriétés semi conductrices (conditionnées par la nature des défauts ponctuels) devrait permettre une meilleure compréhension des mécanismes de formation de ces couches d'oxyde. / The chemical durability of the metal alloy results in particular from the nature of point defects providing transport through the oxidation film formed on the surface. Models oxide layers, grown by thermal oxidation and Alomic Layer Deposition, will be studied by photoelectrochemistry. This will provide us information about the semiconductive properties of the oxide, determined by the point defect in the oxide layer, and should allow us a better understanding of the formation mechanism of these oxide.

Les travaux portent sur deux nuances d’alliages Ni-Cr-W, l’une contenant des précipités de W (EM722) et l’autre non (EM721). Les matériaux forgés, fournis par Aubert & Duval, présentaient des microstructures hétérogènes avec de très gros grains. Dans un premier temps, une procédure thermomécanique a été développée afin d’obtenir un état de référence avec une microstructure homogène avec une faible taille de grains et une texture relativement peu accusée. A partir de cet état de référence, l’évolution de la texture, au cours du laminage à froid (différents taux de réduction), a été analysée par diffraction des neutrons et a montré un renforcement de la texture de type Laiton. Les mécanismes de déformation ont été étudiés en microscopie électronique en transmission et par EBSD (Electron BackScatter Diffraction) afin de comprendre la formation de ce type de texture. Les bandes de cisaillement sont supposées être un facteur important pour la formation de ce type de texture. A partir des états déformés, la cinétique de recristallisation a été déterminée pour deux températures (700°C et 900°C) à partir de mesures de microdureté et par EBSD. Les mécanismes de recristallisation ont été étudiés par MET. Le mécanisme de SIBM (Strain Induced grain Boundary Migration) est le mécanisme prépondérant pour la nuance EM721. Pour l’autre nuance, EM722, possédant des précipités de W, le mécanisme de SIBM est présent, mais en plus, un second mécanisme de PSN (Particle Stimulated Nucleation) a été mis en évidence. De plus, quel que soit le mécanisme de recristallisation, une fois que les germes sont formés, ils maclent très rapidement. Par conséquent, une étude complémentaire sur l’évolution des joints de macle pendant la recristallisation primaire a été réalisée, en employant l’EBSD et surtout le système ASTAR installé dans un MET. Il a alors été montré que cette évolution dépend à la fois de la taille des grains et de l’énergie stockée par les grains pendant la déformation. Enfin, une simulation de la recristallisation primaire de type Monte-Carlo a également été mise en œuvre. Son originalité réside notamment dans sa capacité à rendre compte de la formation des joints de macles cohérents et rectilignes. / This work focuses on two grades of Ni-Cr-W alloys, one containing W precipitates (EM722) and the other not (EM721). The forged materials, provided by Aubert & Duval, had heterogeneous microstructures with very large grains. At first, a thermo-mechanical procedure was developed to obtain a reference state with a homogeneous microstructure with a small grain size and a texture relatively weak. From this reference state, the texture evolution during cold rolling (different reduction rates), was analyzed from neutron diffraction measurements and showed the strengthening of the Brass-type texture. The deformation mechanisms were studied by TEM (transmission electron microscopy) and EBSD (Electron Backscatter Diffraction) to understand the formation of this type of texture. The formation of shear bands is assumed to be the most important factor for the formation of such a texture. From the deformed states, the recrystallization kinetics was determined for two temperatures (700°C and 900°C) from micro-hardness and EBSD measurements. The recrystallization mechanisms were investigated by TEM. The SIBM (Strain Induced Grain Boundary Migration) mechanism is the predominant mechanism for the EM721 material. For the other one (EM722), which has W precipitates, the SIBM mechanism is present, but a second one (PSN: Particle Stimulated Nucleation) has also been evidenced. Furthermore, regardless of the mechanism of recrystallization, once the nuclei are formed, they twin quickly. Therefore, a further study on the evolution of twin boundaries during primary recrystallization was performed, using EBSD and especially the ASTAR system installed in a TEM. It has been shown that the twin development depends both on the grain size and the stored energy during the deformation. Finally, a Monte Carlo simulation of the primary recrystallization was also implemented. Its originality lies especially on its ability to create coherent and straight twin boundaries.

Zr cladding alloys have been used for many years as the first safety barrier layer of a nuclear reactor. However, the recent Fukushima accidents and industrial demands to increase the burnup of fuels have led to increasing interest in a detailed mechanistic understanding of aqueous corrosion and hydrogen pickup and the performance at high temperatures. As part of an international MUZIC-2 programme (Mechanistic Understanding of Zr Corrosion and Hydrogen pickup), I have used a range of advanced microscopy techniques to study the microstructure, the nanoscale chemistry and the porosity in a series of zirconium alloys at different stages of corrosion and hydrogen pickup. Samples from both autoclave and in-reactor conditions were available to compare, I have focussed on RXA (recrystallised 580°C) Zr-1.0Nb and annealed (720°C) Zr-1.0Nb alloys. A set of samples from different exposures times were chosen to represent early, pre-transition and post-transition samples in order to compare the microstructure and microchemistry of the oxides, the metal-oxide interface and the metal. (Scanning) Transmission Electron Microscopy ((S)TEM), Transmission Kikuchi Diffraction (TKD) and automated crystal orientation mapping with TEM (ASTAR mapping) were used to study the grain structure and phase distribution. Significant differences in grain morphology were observed between samples oxidised in the autoclave with different corrosion rates, with more uneven metal-oxide interface, more parallel cracks and less organised oxide grains in the fast corroding samples. Comparing with autoclave samples, the in-reactor samples have shorter, less well-aligned monoclinic grains and more tetragonal grains. The rapidly oxidising annealed Zr-1.0Nb alloy also have much higher tetragonal grain fraction comparing with the slow corrosion rate RXA Zr-1.0Nb alloys. Porosity in the oxide is predicted to have a major influence on the overall rate of corrosion and hydrogen pickup, and there is much more porosity in the annealed Zr-1.0Nb alloy than found in either the RXA alloy or the similar alloy exposed to neutron irradiation. A combination of Energy Dispersion X-ray (EDX) mapping in STEM and Electron Energy Loss Spectroscopy (EELS) analysis of second phase particles can reveal the main and the minor element distributions respectively. The annealed Zr-1.0Nb alloys have Î2-Zr SPPs with nano crystalline structure and much larger size. Although they does not relate with the higher density of cracks in the oxide, the large SPP size can connect together all the small cracks that are generated by the huge amount of tetragonal to monoclinic phase transformation during corrosion and provides pathway for corrosion and hydrogen pickup. Two kinds of SPPs are found in the RXA Zr-1.0Nb alloys, one is Î2-Nb and another one is Zr-Nb-Fe Laves phase. Neutron irradiation seems to have little effect on promoting fast oxidation or dissolution of Î2-Nb precipitates, but encourages dissolution of Fe from Laves phase precipitates. Electron Energy Loss Spectroscopy (EELS) analysis of the oxidation state of Nb in Î2-Nb SPPs in the oxide revealed the fully oxidised Nb⁵⁺ state in the SPPs deep into the oxide, but Nb²⁺ in the crystalline SPPs near the metal-oxide interface. EELS, TKD and ASTAR mapping have also revealed the presence of suboxide layers with the hexagonal ZrO structure predicted by ab initio modelling. The combined thickness of the ZrO suboxide and oxygen-saturated layers at the metal-oxide interface correlates well to the estimated instantaneous oxidation rate, suggesting that the presence of this oxygen- rich zone combining with the part where porosity is not interconnected is the protective oxide that is rate limiting in the key in the transport processes involved in corrosion and hydrogen pickup.

This thesis covers an implementation of an artificial intelligence (AI) system for cooperation between multiple AI-agents. It was done as a part of a master thesis in Media Technology and a master thesis in Computer Engineering at Linköping University, campus Norrköping. The aim of the project was to explore modern techniques in AI and also develop a platform where this AI is implemented for the upcoming educational purposes. The idea is that students can use the system as a base to extend, learn and implement their own AI algorithms. Based on a literature study in AI systems the decision was made to base APAC on the GOAP system, a scalable planning architecture designed for real-time control of autonomous character behavior in games. The result of the thesis is a virtual world, made in Unity3D and C#, where the system is being used by virtual agents to build a city.

Les travaux portent sur deux nuances d'alliages Ni-Cr-W, l'une contenant des précipités de W (EM722) et l'autre non (EM721). Les matériaux forgés, fournis par Aubert & Duval, présentaient des microstructures hétérogènes avec de très gros grains. Dans un premier temps, une procédure thermomécanique a été développée afin d'obtenir un état de référence avec une microstructure homogène avec une faible taille de grains et une texture relativement peu accusée. A partir de cet état de référence, l'évolution de la texture, au cours du laminage à froid (différents taux de réduction), a été analysée par diffraction des neutrons et a montré un renforcement de la texture de type Laiton. Les mécanismes de déformation ont été étudiés en microscopie électronique en transmission et par EBSD (Electron BackScatter Diffraction) afin de comprendre la formation de ce type de texture. Les bandes de cisaillement sont supposées être un facteur important pour la formation de ce type de texture. A partir des états déformés, la cinétique de recristallisation a été déterminée pour deux températures (700°C et 900°C) à partir de mesures de microdureté et par EBSD. Les mécanismes de recristallisation ont été étudiés par MET. Le mécanisme de SIBM (Strain Induced grain Boundary Migration) est le mécanisme prépondérant pour la nuance EM721. Pour l'autre nuance, EM722, possédant des précipités de W, le mécanisme de SIBM est présent, mais en plus, un second mécanisme de PSN (Particle Stimulated Nucleation) a été mis en évidence. De plus, quel que soit le mécanisme de recristallisation, une fois que les germes sont formés, ils maclent très rapidement. Par conséquent, une étude complémentaire sur l'évolution des joints de macle pendant la recristallisation primaire a été réalisée, en employant l'EBSD et surtout le système ASTAR installé dans un MET. Il a alors été montré que cette évolution dépend à la fois de la taille des grains et de l'énergie stockée par les grains pendant la déformation. Enfin, une simulation de la recristallisation primaire de type Monte-Carlo a également été mise en œuvre. Son originalité réside notamment dans sa capacité à rendre compte de la formation des joints de macles cohérents et rectilignes.

This thesis explores and compares traditional and reinforcement learning (RL) methods of performing 2D flight path planning in 3D space. A wide overview of natural, classic, and learning approaches to planning s done in conjunction with a review of some general recurring problems and tradeoffs that appear within planning. This general background then serves as a basis for motivating different possible solutions for this specific problem. These solutions are implemented, together with a testbed inform of a parallelizable simulation environment. This environment makes use of random world generation and physics combined with an aerodynamical model. An A* planner, a local RL planner, and a global RL planner are developed and compared against each other in terms of performance, speed, and general behavior. An autopilot model is also trained and used both to measure flight feasibility and to constrain the planners to followable paths. All planners were partially successful, with the global planner exhibiting the highest overall performance. The RL planners were also found to be more reliable in terms of both speed and followability because of their ability to leave difficult decisions to the autopilot. From this it is concluded that machine learning in general, and reinforcement learning in particular, is a promising future avenue for solving the problem of flight route planning in dangerous environments.