Abstract
Simultaneous localization and mapping (SLAM) is a fundamental problem for various applications. For indoor environments, planes are predominant features that are less affected by measurement noise. In this paper, we propose a novel point-plane SLAM system using RGB-D cameras. First, we extract feature points from RGB images and planes from depth images. Then plane correspondences in the global map can be found using their contours. Considering the limited size of real planes, we exploit constraints of plane edges. In general, a plane edge is an intersecting line of two perpendicular planes. Therefore, instead of line-based constraints, we calculate and generate supposed perpendicular planes from edge lines, resulting in more plane observations and constraints to reduce estimation errors. To exploit the orthogonal structure in indoor environments, we also add structural (parallel or perpendicular) constraints of planes. Finally, we construct a factor graph using all of these features. The cost functions are minimized to estimate camera poses and global map. We test our proposed system on public RGB-D benchmarks, demonstrating its robust and accurate pose estimation results, compared with other state-of-the-art SLAM systems.
Highlights
Simultaneous localization and mapping (SLAM) develops quickly in recent years and becomes a fundamental problem for various applications including mobile robots, augmented and virtual reality
Like other modern SLAM systems, ours can be divided into two functional parts: (1) frond-end, the tracking part extracts and matches features for new captured frame, and estimates the camera pose by minimizing the error function constituted by the tracked features in the map; (2) back-end, the map management part estimates and optimizes landmarks in the environment
We propose a novel point-plane SLAM system for indoor environments
Summary
Simultaneous localization and mapping (SLAM) develops quickly in recent years and becomes a fundamental problem for various applications including mobile robots, augmented and virtual reality. Indoor environments are common working scenes for mobile robots These high-level features ensure faster and more accurate data association, which can be extracted using RGB-D cameras. The planes calculating from many points are more robust and accurate, because of less affection from measurement noise Using these high-level features helps to improve the performance of SLAM. There are various man-made objects and structures, which have lots of parallel and perpendicular planes Using these kinds of structural constraints can help to achieve a long-term association for planes, resulting in smaller accumulated error. The real planes in working environments have limited size, contours or edges These features can be exploited to add constraints for robust pose estimation. We evaluate our proposed system on public datasets and achieve state-of-the-art performance, which performs nearly in real time
Sign-in/Register to view highlights & summary 
Published Version (
Free)
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call