| This thesis addresses the problem of the design, development, and validation of an integrated autonomous navigation system for a mobile robot exploring and navigating in an unknown or partially known drift-type environments. Underground mines or caves are typical examples of such environment. In this context, we propose a system allowing two modes of operation: exploration/surveying mode and autonomous navigation mode.; In the exploration/surveying mode, the supervisor specifies a motion path through designated areas on a preliminary sketch of the environment provided on-line by sensors, using a remote interface. Geometrical data from range finder, odometry, and inertial sensors are then collected along the path and concatenated into 2D/3D metric maps of the environment.; In the autonomous navigation mode, high-level missions are specified based on the previously acquired 2D map. A motion planner translates each mission into a set of consecutive navigation actions; each one is delimited by natural landmarks (such as intersections) and consists of a navigation path and a navigation mode. A map-based navigation executive controller executes navigation actions autonomously. Mission execution involves a combination of functions: multisensor data acquisition, processing and integration, odometry and wall-guided motion control, path planning and tracking control, obstacles detection, sensing and surveying, and 2D/3D maps building, vehicle localization, and natural landmarks detection.; Each function mentioned above is developed and integrated in an experimental platform which has been successfully tested in the laboratory and in an underground mine. This thesis describes the platform, the system architecture, the task execution algorithms structures, and it also presents the results of the experiments.; The innovative contribution of this thesis can be summarized as follow: (1) Design and development of an executive controller for map-based navigation. (2) Development of sensor integration strategies to integrate data from the gyroscope, inclinometer and odometry. (3) Development of a self-localization method based on natural landmarks (intersections and bays) identification. (4) Development of an autonomous 2D/3D surveying system for an unstructured environment. (Abstract shortened by UMI.)... |
