The current state-of-the-art in Autonomous Ground Vehicle (AGV) technology requires expensive, delicate laser range finders to apperceive the environmental impact of driving. The situation of too costly ladar represents a large barrier to adoption of AGV in the future, whereas provides an opportunity for closeto-market large range sonar sensor.In this paper, we propose an obstacle detection algorithm using adjacent periods' echo data of the large range sonar sensor in the off-road environment. We first integrate vehicle odometry into the sonar sensor and succeed in changing one dimension (1D) distance information into two dimension (2D) signal, which provides a strong prior constraint to filter unstable noisy echoes. We use Hungarian algorithm to solve correspondence of data points to make sure they are reflected back by a mutual object. Matched dual points are used to extract the obstacle's line feature represented in the manner of common tangent of the two intersecting arcs. Experiments in outdoor environment demonstrate validity of our algorithm.
Scan registration plays a critical role in odometry, mapping and localization for Autonomous Ground Vehicle. In this paper, we propose to adopt a probabilistic framework to model the locally planar patch distributions of candidate points from two or more consecutive scans instead of the original point-to-point mode. This can be regarded as the plain-to-plain measurement metric which ensures a very high confidence in the normal orientation of aligned patches. We take into account the geometric attribution of the scanning beam to pick out feature points and then which can reduce the number of selected points to a lower level. The optimization of transform is achieved by the combination of high frequency but coarse scan-to-scan motion estimation and low frequency but fine scan-to-map batch adjustment. We validate the effectiveness of our method by the qualitative tests on our collected point clouds and the quantitative comparisons on the public KITTI odometry datasets.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.