Simultaneous Localization And Mapping Based On B-Spline Feature

In order to successfully accomplish all the tasks in an unknown environment, themobile robot must have the ability to achieve autonomous navigation by using its ownposition and the maps of surrounding environment. The estimation problem for robotpose and environmental feature map is usually referred to the simultaneouslocalization and mapping (SLAM). As SLAM has increasingly become a hotspot forrobot applications, the feature extraction in SLAM is also drawing more and moreattention. There are several methods for feature extraction, while methods based onpoint-feature extraction and line-feature extraction are more commonly used thanothers. The two solutions have been successfully applied in SLAM problem, but asthe presence of curved geometric features in modern constructions, it becomes theurgent problem of feature extraction to accurately and efficiently describe thegradually increasing complexity of the environment.Taking into account the complexity of the environmental features and thelimitations of using point features and line features to describe the environment,B-Splines SLAM, a novel feature-based solution to the SLAM problem, based on theutilization of B-spline curves to represent the features in complex environments, ispresented. Features are described using B-splines as modeling tool, and the set ofcontrol points defining their shape is used to form a complete and compact descriptionof the environment, thus making it feasible to use an extended kalman filter(EKF).The power and computational efficiency of B-spline curves are combined with anEKF-SLAM framework, thus allowing the representation of complex structures in aparametric way. First, the introduction provides a brief overview of the backgroundand the significance of this research, the classification of the mobile robot navigationtechnology and their advantages and disadvantages, the history and valuable results ofthe SLAM algorithm. Secondly, fundamental theories regarding the SLAM algorithmare presented, such as the system states, the vehicle and landmark models, as well asthe general structure framework of EKF-SLAM and the mathematical formulas ofeach part. Thirdly, the basic concepts of B-splines are provided including its definition,characteristics, the method for spline fitting and the computation of control points. More importantly, extensive experimental results are showed to verify the impact ofthe order and the knots spacing of the B-splines on spline fitting. And then theB-spline theory and the SLAM algorithm are organically blend together to form theB-splines SLAM algorithm which is the core content of this article. The processes offeature extraction, data association, prediction, observation, updatation and mapenlargement are introduced in detail. Initial simulation experimental result indicatesthat the B-splines SLAM is feasible and effective regardless of the geometric complexof the environment. Finally, the conclusion and further improvement are presented.