A multi-feature localization algorithm for mobile robots indoor environment mapping

Oriented to meet the needs of environmental modeling of indoor service robots and to facilitate the indoor environmental feature extraction, this paper proposes an indoor feature fusion approach through which the problem of the stability of the environmental feature identification is solved, and the feature map is simplified. In this approach, we first fit the segment feature and the circle feature with a constrained least square method, and then, we put forward a feature fusion method which respectively fuses the segment feature and the circle feature, taking Hotelling T2 hypothesis testing as basis to test which features can be fused, and further we use the measurement of feature as weight to fuse those features. The results of the experiment indicated that the method can stably fuse environmental features and simplify the feature map in the indoor environment.