Cette thèse de doctorat s’inscrit dans le cadre des activités de recherche sur les systèmes de lancement aéroporté autonome. L’originalité du travail est basée sur la planification de mission effectuée par un algorithme de type A*(A-étoile). Cet algorithme a été amélioré pour répondre aux besoins de la mission de largage d’un lanceur. Il effectue la planification du chemin le plus court dans un espace tridimensionnel. Le meilleur chemin est choisi à partir de plusieurs points de passage générés dans la région de mission. Une région peut être une phase du vol ou une partie du profil de vol. Le chemin le plus court est identifié par rapport à la présence de différents obstacles dans l’espace de recherche et son objectif consiste à atteindre un point désiré. Les obstacles ont différentes dimensions et orientations dans l’espace. L’étude de leur comportement est associée aux incertitudes en provenance de l’environnement. Ils peuvent représenter des régions interdites au vol ou des conditions atmosphériques défavorables. L’évolution de ces derniers n’est pas prévisible à l’avance, ce qui impose l’addition d’une fonctionnalité dans l’algorithme. Il est possible de replanifier le chemin à partir d’un point de passage appartenant à un chemin généré en fonction de la position détectée récemment de l’obstacle en déplacement pour arriver dans la configuration finale désirée. Cette détection est possible grâce aux capteurs positionnés sur le premier étage de ce système de lancement représenté par un avion-porteur. Les points de passage que le véhicule aérien doit suivre pour atteindre les objectifs importants ne sont pas choisis d’une manière aléatoire. Leur génération dans l’espace de recherche du chemin est définie en rapport aux limitations dynamiques de l’avion. Les modèles cinématique et dynamique du véhicule aérien qui décrivent son évolution sont aussi développés dans cette thèse. Ces modèles sont étudiés dans un système de coordonnées aérodynamiques. Le référentiel traite la présence du vent qui influe sur le comportement du véhicule. Cela nous permet de considérer d’une manière prédictive plusieurs incertitudes en provenance de l’environnement ou internes pour le véhicule. Les perturbations internes sont provoquées par le largage du lanceur. Le régime transitoire est relié à la perte de masse qui pour certaines missions peut atteindre le tiers de la masse totale du système de lancement. L’algorithme de planification traite une autre prévision – la possibilité que le largage ne soit pas réalisé. Cela peut arriver dans le cas où une tempête s’est installée dans la région de lancement ou il y a plusieurs obstacles dont l’évitement risque de consommer trop de carburant et d’empêcher le retour sur le site d’atterrissage. Les connexions entre les différents points de passage peuvent être souvent brutes et difficiles à réaliser par le véhicule aérien. Pour résoudre cette problématique dans le deuxième module développé sur la génération de trajectoire réalisable, nous utilisons l’approche des polynômes de troisième ordre. Ces polynômes par rapport aux autres techniques diminuent le temps du calcul pour générer une trajectoire réalisable entre deux points de passage consécutifs. Le chemin réalisable est facile à suivre par le système. Pour le suivi de la trajectoire, nous avons introduit dans un troisième module – la commande par mode glissant. Le principe de cette commande consiste le choix de la surface de commutation entre la trajectoire actuelle suivie par le véhicule et la trajectoire désirée déterminée par l’algorithme de planification A-étoile et générée par les polynômes cartésiens de troisième ordre. / This Ph.D. thesis deals with the systems of autonomous airborne launch vehicles. The originality of this work is based on the mission planning released by a graph-based A* (A-star) pathfinding algorithm. This algorithm was improved to respond to the specifications of this launching mission. It carries out the planning of the shortest path in a three-dimensional space. The optimal path is selected from the interconnections of several waypoints generated in the mission area. An area can be a specific mission phase or a part of the flight plan. The shortest path is identified according to the presence of various obstacles during the path search and its objective is to reach a desired point in the region. The obstacles have various dimensions and orientations in space. The study of their behavior is associated with disturbances coming from the environment. They could be forbidden flight regions or unfavorable atmospheric conditions. The evolution of the latter cannot be always predicted in advance, which still imposes an improvement that can be added in the operation of the algorithm. The path replanning is also possible. Starting from a safe waypoint from an already generated path according to a recently detected obstacle, a new path can be planned from this point considering the new obstacle coordinates to arrive at the desired final configuration. This detection will be taken into account by the sensors situated on the airborne launcher called a carrier to define the final necessary computing time. The waypoints which the airborne vehicle must follow to achieve the important mission goals are not selected in a random manner. Their generation in the search space is defined according to the dynamic limitations of the vehicle. The kinematic and dynamic models of the carrier are also developed in this thesis. These models are studied in an aerodynamic reference frame. This frame treats the presence of the wind which influences the vehicle evolution in space. That enables to consider in a predictive manner several uncertainties coming from the environment or internal for the vehicle. The internal disturbances are caused by the launching mode relied to a significant loss of mass which for certain missions can reach a half of the total mass of the launching system. The planning algorithm treats in a predictive manner – the possibility that the launching is not executed. That can happen if in the launching region a storm is settled or there are several obstacles that avoidance is likely to consume the fuel of the carrier and to prevent the successful return on the landing site. The interconnections between the various waypoints can be often rough and difficult to execute by the airborne launcher. To solve these problems a second module has to be developed to generate a feasible trajectory using the polynomials of third order.. Compared to other techniques this one decreases the calculation time of the trajectory between two consecutive waypoints. The feasible path is easy to follow by the airborne launcher. For the trajectory tracking we introduced into a third module the sliding mode control. The functionality of this control is in the choice of switching surfaces between the current trajectory tracking by the vehicle and the desired trajectory defined by the A* algorithm waypoints and generated by the third order polynomials.

No description available.

In today's digital (electronics) world, people's desire for electronic goods that ease their life at work, and leisure is increasing the complexity of the products of the embedded systems industry. For example, MP3 players for listening to music and cell phones for communicating with people.The gap between the hardware and software parts of embedded systems is being reduced by the use of System Level Design Languages (SLDL) that can model both hardware and software simultaneously. One such SLDL is SpecC.In this thesis, a SpecC model of a Real Time Operating System (RTOS) is constructed. It is shown how RTOS features can be incorporated into a SpecC model. The model is used to develop an application involving a robot avoiding obstacles to reach its destination. The RTOS model operates similar to the actual RTOS in the robot.The application includes a testbench model for the robot, including features such as interrupts, sonar sensors and wheel pulses, so that its operation closely resembles the actual robot. The sensor model is programmed to generate the values from the four sensor receivers, similar to the behaviour of the sensors on the actual robot. Also the pulses from the wheels and associated interrupts are programmed in the model so that it resembles the interrupts and wheel pulses present on actual robot.

This thesis presents an indoor obstacle avoidance system for car-like mobile robot. The system consists of stereo vision, map building, and path planning. Stereo vision is performed on stereo images to create a geometric map of the environment. A fast sparse stereo approach is employed. For different areas of the image there are different optimal values of disparity range. A multi-pass method to combine results at different disparity range is proposed. To reduce computational complexity the matching is limited to areas that are likely to generate useful data. The stereo vision system outputs a more complete disparity map. Abstract Map building involves converting the disparity map into map coordinates using triangulation and generating a list of obstacles. Occupancy grids are built to aid a hierarchical collision detection. The fast collision detection method is used by the path planner. Abstract A steering set path planner calculates a path that can be directly used by a car-like mobile robot. An adaptive approach using occupancy grid information is proposed to improve efficiency. Using a non-fixed steering set the path planner spends less computation time in areas away from obstacles. The path planner populates a discrete tree to generate a smooth path. Two tree population methods were trialled to execute the path planner. The methods are implemented and experimented on a real car-like mobile robot.

In this master thesis a control strategy for a Human Power Amplifier (HPA) ispresented. An HPA can be described as a machine that amplifies a force exertedby a human operator. The HPA in this thesis can best be described as a mechanicalore with two degrees of freedom.The approach for the control strategy presented here is to look at the controlproblem not directly as a force amplifying problem, but as coordination problembetween the real system and a virtual system, where the virtual system is used asa reference. If the systems are synchronized then desired force amplification willnaturally follow from that.Furthermore is the possibility to implement guidance and obstacle avoidanceon the machine investigated. The guidance is performed by using velocity fields,i.e., vector fields where a vector represents the desired velocity for each point inthe plane. For the obstacle avoidance potential fields are used, where the idea isthat a high potential should repel the machine from restricted areas.

This thesis describes the development of the Texas A&M University Autonomous Helicopter System and the integration of obstacle avoidance capabilities into that system. The helicopter system, composed of a Bergen Observer helicopter and a Rotomotion Autonomous Flight Control System (AFCS), was developed as a platform to support the development of the obstacle avoidance system through integration of sensors and onboard processing capabilities. The system has proven in various flight tests that it has the capability to autonomously hover and fly to user defined GPS waypoints. The obstacle avoidance algorithm has been proven in simulations involving an interface with the Rotomotion AFCS and the flight simulation software they created to facilitate the development of that system. The helicopter has also demonstrated appropriate responses to sensor input commensurate with the obstacle avoidance algorithm. Full avoidance tests were unable to be performed due to hardware malfunctions inherent in the obstacle avoidance sensors.

This thesis presents an indoor obstacle avoidance system for car-like mobile robot. The system consists of stereo vision, map building, and path planning. Stereo vision is performed on stereo images to create a geometric map of the environment. A fast sparse stereo approach is employed. For different areas of the image there are different optimal values of disparity range. A multi-pass method to combine results at different disparity range is proposed. To reduce computational complexity the matching is limited to areas that are likely to generate useful data. The stereo vision system outputs a more complete disparity map. Abstract Map building involves converting the disparity map into map coordinates using triangulation and generating a list of obstacles. Occupancy grids are built to aid a hierarchical collision detection. The fast collision detection method is used by the path planner. Abstract A steering set path planner calculates a path that can be directly used by a car-like mobile robot. An adaptive approach using occupancy grid information is proposed to improve efficiency. Using a non-fixed steering set the path planner spends less computation time in areas away from obstacles. The path planner populates a discrete tree to generate a smooth path. Two tree population methods were trialled to execute the path planner. The methods are implemented and experimented on a real car-like mobile robot.

In today's digital (electronics) world, people's desire for electronic goods that ease their life at work, and leisure is increasing the complexity of the products of the embedded systems industry. For example, MP3 players for listening to music and cell phones for communicating with people.The gap between the hardware and software parts of embedded systems is being reduced by the use of System Level Design Languages (SLDL) that can model both hardware and software simultaneously. One such SLDL is SpecC.In this thesis, a SpecC model of a Real Time Operating System (RTOS) is constructed. It is shown how RTOS features can be incorporated into a SpecC model. The model is used to develop an application involving a robot avoiding obstacles to reach its destination. The RTOS model operates similar to the actual RTOS in the robot.The application includes a testbench model for the robot, including features such as interrupts, sonar sensors and wheel pulses, so that its operation closely resembles the actual robot. The sensor model is programmed to generate the values from the four sensor receivers, similar to the behaviour of the sensors on the actual robot. Also the pulses from the wheels and associated interrupts are programmed in the model so that it resembles the interrupts and wheel pulses present on actual robot.

No description available.

This project is carried out in the field of motion planning of AGVs for the company AGVE with the aim to automate the obstacle free trajectory generation process. The process of manually generating splines in AutoCAD to achieve obstacle avoidance is replaced by the automatic generation of paths by running a python script. Artificial potential field algorithm is implemented in the python script to achieve obstacle avoidance. Clothoid curve is used to create feasible trajectories for obtaining obstacle free paths. The developed program is tested and proved suitable on three scenarios including a real-life problem encountered by the company. The output obstacle free path can be manipulated using three factors which are alpha, moving the neighbours and manipulating the scaling factor for potential fields in the APF algorithm.