This paper presents a customized FastSLAM algorithm that takes both corners and line segments as map features to perform simultaneous localization and mapping. Unlike the conventional SLAM or FastSLAM algorithms, which only extract corner or point features and have a biased priority given to localization rather than mapping, the line segment features in this new algorithm naturally form a meaningful map of the indoor environment. In addition, using a variety of features is believed to result in a more robust performance. The proposed algorithm is tested based on measurements from a Hokuyo scanning laser range finder mounted on an quadrotor UAV developed by the UAV Research Team from the National University of Singapore.