The annotated laser data set for navigation in urban areas

Abstract

In this paper, we present a challenging data set for use in a variety of mobile robotic navigation tasks. Based on the data set that we released previously, we further extend its capability by annotating the laser range data. Apart from the laser range data, the dead-reckoning trajectory and the tri-camera imagery, the true segmentation differentiating between objects and the true moving object detection (MOD) discriminating moving objects are also annotated. The data set is intended for extensive evaluation of localization, mapping and MOD by providing highly accurate relative pose, object segmentation and MOD. All data are synchronized, carefully annotated and formatted in space-delimited plaintext format. We also provide an interactive web interface to facilitate the access to the data and the annotations.