A Forward-Viewing Mono-Camera Based SLAM System for Indoor Service Robots

Abstract

This dissertation presents a new forward-viewing monocular vision-based simultaneous localization and mapping (SLAM) method. The method is developed to be applicable in real-time on a low-cost embedded system for indoor service robots. The developed system utilizes a cost-effective mono-camera as a primary sensor and robot wheel encoders as well as a gyroscope as supplementary sensors. The proposed method is robust in various challenging indoor environments which contain low-textured areas, moving people, or changing environments. In this work, vanishing point (VP) and line features are utilized as landmarks for SLAM. The orientation of a robot is directly estimated using the direction of the VP. Then the estimation models for the robot position and the line landmark are derived as simple linear equations. Using these models, the camera poses and landmark positions are efficiently corrected by a novel local map correction method. To achieve high accuracy in a long-term exploration, a probabilistic loop detection procedure and a pose correction procedure are performed when the robot revisits the previously mapped areas. The performance of the proposed method is demonstrated under various challenging environments using dataset-based experiments using a desktop computer and real-time experiments using a low-cost embedded system. The experimental environments include a real home-like setting and a dedicated Vicon motion-tracking systems equipped space. These conditions contain low-textured areas, moving people, or changing environments. The proposed method is also tested using the RAWSEEDS benchmark dataset.