Research On Autonomous Navigation Algorithm Of Mobile Robot Based On The Characteristics Of Curve Modeling

So far, many studies on simultaneous localization and mapping (SLAM)algorithm use point as the feature of map, that is the map is composed of a finitenumber of points. The biggest advantage of using point features is the simple andintuitive calculation, and point-based SLAM is basically mature. With the needs of theactual situation, line features appear later, and has been successfully applied in thedescription of indoor environment. Point features and the line features can be wellapplied in structured environment, but in the unstructured environment, the situationis always complex—the shape is irregular and there is no obvious edge, so it cannoteffectively extract point or line features to describe the environment. This paper putforward a method that could be applied to the unstructured environment—B spline(also called basic spline).BS_SLAM method uses B spline function as the basis function, a set of controlpoints to represent a spline, and a set of node vector to determine the location of thespline. Currently, the theory of B spline curve has been mature, and variousalgorithms and parameters have been systematized, such as the algorithm ofdetermining the node vector with the Hartley-Judd method, the curve extensionalgorithm proposed by Shetty and White, the de Boor algorithm of calculating a valueof a point on the spline curve，etc. The significant advantage of using B spline todescribe the environment is that B spline features contains point and (straight) linefeatures, but not limited to these two features, and the low environment requirements.BS_SLAM has three following characteristics: First, people cannot directly observethe control points, they can only obtain them from the data points; secondly, the dataassociation between splines is far more complex than single point in theory; thirdly,the point features is discrete, but the B-Spline curve is continuous. In the method ofBS_SLAM, sometimes spline extension is necessary after data association, and it willincrease the complexity of calculation.There are two innovations in this paper: First, it realizes the intelligent selectionof control points. This article uses dynamic node vectors to describe the splines, anddetermine the number of control points according to the number of data points.Second, it proposes a new data association algorithm, including the initial associationand the confirm association. According to the nearest distances principle, it makes thesplines in initial association simple and uses the simplified map to confirm the finalassociation.