We build a simultaneous localization and mapping (SLAM) system based on Microsoft Kinect 2 and Robot Operating System (ROS). Our system can estimate the trajectory of a hand held Kinect and generate the point cloud or octomap with color information of the environment. We present our approaches and system architecture in this paper and evaluate its performance through some published datasets as well as data from Kinect. The experiments demonstrate that our system is robust under slow sensor motion and relatively simple environment.