Robust and High Accuracy Autonomous Navigation

Book excerpt: Autonomous navigation of a mobile vehicle is a challenging task. almost all applications of an autonomous vehicle require a vehicle moving accurately and repeatedly to a particular location within its environment, while it is executing a specific task, without continuous human guidance. The goal of this dissertation is to develop new motion models and new algorithms, to improve the accuracy and the robustness and solve conflicting situations regarding landmark based autonomous navigation. Therefore, this dissertation is proposing a new composite local speed motion model for a local 3D environment and a new composite local odometry-inertial motion model for a local 3D environment. Efficient algorithms that can guarantee real-time performance in the presence of insufficient or conflicting data are proposed as well. These algorithms are solving problems like: lost situations, false readings, multiple reflections, or independent distance measurements. As modeling the contents of the immediate environment is fundamental, estimation of the vehicle position with respect to the external world is fundamental also. The task for this dissertation will be completed by using the most powerful tools of estimation theory in noisy environments: extended and unscented Kalman filtering to estimate the autonomous vehicle's position.