An efficient approach of map-learning on service robot in complex office environment using laser radar

This paper presents an efficient approach to simultaneous localization and mapping problems (SLAM) in complex office environment for a service robot without the capability of self-localization but equipped with a 2D laser radar. We propose a model of multi-layer bipartite graph of boundary segments extracted from successive data frames. To eliminate the cumulative errors caused by dead reckoning and radar itself, we introduce an error vector model defined between two successive layers and correct each data frame based on error vectors. The experimental results in a real office environment show that the approach we proposed can effectively reduce the cumulative errors during localization and mapping.