Visual Navigation Using Heterogeneous Landmarks and Unsupervised Geometric Constraints

Abstract

We present a heterogeneous landmark-based visual navigation approach for a monocular mobile robot. We utilize heterogeneous visual features, such as points, line segments, lines, planes, and vanishing points, and their inner geometric constraints managed by a novel multilayer feature graph (MFG). Our method extends the local bundle adjustment-based visual simultaneous localization and mapping (SLAM) framework by explicitly exploiting the heterogeneous features and their inner geometric relationships in an unsupervised manner. As the result, our heterogeneous landmark-based visual navigation algorithm takes a video stream as input, initializes and iteratively updates MFG based on extracted key frames, and refines robot localization and MFG landmarks through the process. We present pseudocode for the algorithm and analyze its complexity. We have evaluated our method and compared it with state-of-the-art point landmark-based visual SLAM methods using multiple indoor and outdoor datasets. In particular, on the KITTI dataset, our method reduces the translational error by 52.5% under urban sequences where rectilinear structures dominate the scene.

A Visual-Aided Inertial Navigation and Mapping System

“…Currently, there are two main approaches to implementing vision-based SLAM systems: i) filtering-based methods (see [20][21][22] and [23]), and ii) optimization-based methods (see [24] and [25]). While the former approach is shown to give accurate results when the availability of computational power is enough, filtering-based SLAM methods might still be beneficial if limited processing power is available [26].…”

Robustness to lighting variations: An RGB-D indoor visual odometry using line segments

“…We have studied appearancebased [28]- [30], vertical line-based [14], and heterogeneous features-based [31]- [34] visual navigation. In the process, we have learned the limitations of RGB cameras and the importance of robustness, which leads to this work.…”

Visual‐inertial curve simultaneous localization and mapping: Creating a sparse structured world without feature points

“…As simultaneous localization and mapping (SLAM) has matured as a discipline, SLAM research has increasingly focused on systems‐level issues such as optimizing constituent components of algorithms and overall systems integration . While in the early days of SLAM researchers often presented the results of robot excursions measured in meters, today systems are expected to perform well over much longer trajectories, further emphasizing the need for a systems‐level approach.…”

“…Various SLAM algorithms have incorporated high‐level features in order to overcome drawbacks associated with point‐based SLAM. Examples of high‐level features include planes, image moments, line segments, objects such as office chairs and tables, or a river . A desirable characteristic of high‐level structure is that it provides a compact structured map of the environment.…”

“…The work in Ref. presents a systems‐level SLAM approach that uses line segments, ideal lines, vanishing points, and primary planes that are used in conjunction with feature points. In addition to providing a structured map, their algorithm demonstrates that high‐level features can improve the localization accuracy of a SLAM algorithm when used in conjunction with sparse feature points, and they demonstrate that high‐level features are able to operate in settings that lack distinguishable feature points.…”