Dans la plupart des applications de robotique, un système de vision apporte une amélioration significative de la perception de l’environnement. La vision panoramique est particulièrement intéressante car elle rend possible une perception omnidirectionnelle. Elle est cependant rarement utilisée en pratique à cause des limitations technologiques découlant des méthodes la permettant. La grande majorité de ces méthodes associent des caméras, des miroirs, des grands angles et des systèmes rotatifs ensembles pour créer des champs de vision élargis. Les principaux défauts de ces méthodes sont les importantes distorsions des images et l’hétérogénéité de la résolution. Certaines autres méthodes permettant des résolutions homogènes, prodiguent un flot de données très important qui est difficile à traiter en temps réel et sont soit trop lents soit manquent de précision. Pour résoudre ces problèmes, nous proposons la réalisation d’une caméra panoramique intelligente qui présente plusieurs améliorations technologiques par rapport aux autres caméras linéaires rotatives. Cette caméra capture des panoramas cylindriques homogènes avec une résolution de 6600 × 2048 pixels. La synchronisation de la capture avec la position angulaire est possible grâce à une plateforme rotative de précision. Nous proposons aussi une solution au problème que pose le gros flot de données avec l’implémentation d’un extracteur de primitives qui sélectionne uniquement les primitives invariantes des images pour donner un système panoramique de vision qui ne transmet que les données pertinentes. Le système a été modélisé et une méthode de calibrage spécifiquement conçue pour les systèmes cylindriques rotatifs est présentée. Enfin, une application de localisation et de reconstruction 3D est décrite pour montrer une utilisation pratique dans une application de type Simultaneous Localization And Mapping ( SLAM ). / In most robotic applications, vision systems can significantly improve the perception of the environment. The panoramic view has particular attractions because it allows omnidirectional perception. However, it is rarely used because the methods that provide panoramic views also have significant drawbacks. Most of these omnidirectional vision systems involve the combination of a matrix camera and a mirror, rotating matrix cameras or a wide angle lens. The major drawbacks of this type of sensors are in the great distortions of the images and the heterogeneity of the resolution. Some other methods, while providing homogeneous resolutions, also provide a huge data flow that is difficult to process in real time and are either too slow or lacking in precision. To address these problems, we propose a smart panoramic vision system that presents technological improvements over rotating linear sensor methods. It allows homogeneous 360 degree cylindrical imaging with a resolution of 6600 × 2048 pixels and a precision turntable to synchronize position with acquisition. We also propose a solution to the bandwidth problem with the implementation of a feature etractor that selects only the invariant feaures of the image in such a way that the camera produces a panoramic view at high speed while delivering only relevant information. A general geometric model has been developped has been developped to describe the image formation process and a caligration method specially designed for this kind of sensor is presented. Finally, localisation and structure from motion experiments are described to show a practical use of the system in SLAM applications.

In this thesis, catadioptric omnidirectional imaging systems are analyzed in detail. Omnidirectional image (ODI) formation characteristics of different camera-mirror configurations are examined and geometrical relations for panoramic and perspective image generation with common mirror types are summarized. A method is developed to determine the unknown parameters of a hyperboloidal-mirrored system using the world coordinates of a set of points and their corresponding image points on the ODI. A linear relation between the parameters of the hyperboloidal mirror is determined as well. Conducted research and findings are instrumental for calibration of such imaging systems. The resolution problem due to the up-sampling while transferring the pixels from ODI to the panoramic image is defined. Enhancing effects of standard interpolation methods on the panoramic images are analyzed and edge detection-based techniques are developed for improving the resolutional quality of the panoramic images. Also, the projection surface alternatives of generating panoramic images are evaluated.

This thesis addresses the pose stabilization of a car-like vehicle using omnidirectional visual feedback. The presented method allows a vehicle to servo to a pre-learnt target pose based on feature bearing angle and range discrepancies between the vehicle's current view of the environment and that seen at the learnt location. The best example of such a task is the use of visual feedback for autonomous parallel-parking of an automobile. Much of the existing work in pose stabilization is highly theoretical in nature with few examples of implementations on 'real' vehicles, let alone vehicles representative of those found in industry. The work in this thesis develops a suitable test platform and implements vision-based pose stabilization techniques. Many of the existing techniques were found to fail due to vehicle steering and velocity loop dynamics, and more significantly, with steering input saturation. A technique which does cope with the characteristics of 'real' vehicles is to divide the task into predefined stages, essentially dividing the state space into sub-manifolds. For a car-like vehicle, the strategy used is to stabilize the vehicle to the line which has the correct orientation and contains the target location. Once on the line, the vehicle then servos to the desired pose. This strategy can accommodate velocity and steering loop dynamics, and input saturation. It can also allow the use of linear control techniques for system analysis and tuning of control gains. To perform pose stabilization, good estimates of vehicle pose are required. A simple, yet robust, method derived from the visual homing literature is to sum the range vectors to all the landmarks in the workspace and divide by the total number of landmarks--the Improved Average Landmark Vector. By subtracting the IALV at the target location from the currently calculated IALV, an estimate of vehicle pose is obtained. In this work, views of the world are provided by an omnidirectional camera, while a magnetic compass provides a reference direction. The landmarks used are red road cones which are segmented from the omnidirectional colour images using a pre-learnt, two-dimensional lookup table of their colour profile. Range to each landmark is estimated using a model of the optics of the system, based on a flat-Earth assumption. A linked-list based method is used to filter the landmarks over time. Complementary filtering techniques, which combine the vision data with vehicle odometry, are used to improve the quality of the measurements.