Indoor slope and edge detection by using two-dimensional EKF-SLAM with orthogonal assumption

In an indoor environment, slope and edge detection is an important problem in simultaneous localization and mapping (SLAM), which is a basic requirement for mobile robot autonomous navigation. Slope detection allows the robot to find areas that are more traversable while the edge detection can prevent robot from falling. Three-dimensional (3D) solutions usually require a large memory and high computational costs. This study proposes an efficient two-dimensional (2D) solution to combine slope and edge detection with a line-segment-based extended Kalman filter SLAM (EKF-SLAM) in a structured indoor area. The robot is designed to use two fixed 2D laser range finders (LRFs) to perform horizontal and vertical scans. With local area orthogonal assumption, the slope and edge are modelled into line segments swiftly from each vertical scan, and then are merged into the EKF-SLAM framework. The EKF-SLAM framework features an optional prediction model that can automatically decide whether the application of iterative closest point (ICP) is necessary to compensate for the dead reckoning error. The experimental results demonstrate that the proposed algorithm is capable of building an accurate 2D map swiftly, which contains crucial information of the edge and slope.