An improved algorithm for fast generating maps

In this paper we present an improved algorithm for fast generating maps that combines grid-based FastSLAM with an improved Levenberg-Marquardt iterated closest point (ILM-ICP) scan matching. This approach has some advantages than previous approaches. The ILM-ICP scan matching with much faster than gradient descent is used for efficient minimizing odometric errors during mapping and reducing the number of particles required during resampling step. Simultaneously we propose that the algorithm not only ease the problem of particle depletion but also efficiently decrease the map generating time. Experimental results in three kinds of indoor environment illustrate the advantages of our methods over previous approaches with respect to ability of fast generating maps in real time.