Loop-Closure Detection in Urban Scenes for Autonomous Robot Navigation

Abstract

Relocalization is a vital process for autonomous robot navigation, typically running in the background of sequential localization and mapping to detect loops in the robot's trajectory. Such loop-closure detections enable corrections for drift accumulated during the estimation processes and even recovery from complete localization failures. In this work, we present a novel approach loosely integrated with a keyframe-based SLAM system to perform loop-closure detection in urban scenarios for autonomous robot navigation. Generating a mesh of the current robot's surroundings in real-time using monocular and inertial cues, the proposed method estimates the most salient plane in the current view, enabling the creation of the corresponding orthophoto for this plane. Evaluating image similarity on orthophotos forms a much better conditioned problem for relocalization, minimizing effects from viewpoint changes. Employing binary image descriptors and tests on their relative constellation in the image, the proposed approach exhibits robustness also to illumination and situational variations common in real scenes, overall resulting to significant improvement in loop-closure detection performance in urban scenes with respect to the state of the art.Video -https://www.youtube.com/c/V4RLteam

Viewpoint-Tolerant Place Recognition Combining 2D and 3D Information for UAV Navigation

“…Examples are shown in Figure 4. These sequences were already successfully applied in a place recognition scenario in our previous work [33].…”

Real-Time Wide-Baseline Place Recognition Using Depth Completion

“…However, their increased invariance comes at a prohibitively high computational cost of two orders of magnitude slower than SIFT. By generating a mesh of the current robot's surroundings, the work in [13], makes use of a 3D map provided by SLAM and identifies the most prominent plane in each image computing only one affine transformation, as orthophoto. This enables the creation of a single view of the scene, while using a computationally cheap binary descriptor and avoiding the need for computing multiple transformations of the same image.…”

“…While very efficient in removing outliers, the use of a sparse 3D map together with the local planarity assumption create a smooth mesh of the environment, eliminating details in small areas with a large depth variation. However, as demonstrated in [13] and [24], this approach was already proven to work well in man-made environments, where locally planar structures are usually present. Besides this, this mesh generation approach takes about 7 ms per frame to create a 3D mesh out of the 3D landmarks, rendering it suitable for real-time applications.…”

Visual Place Recognition for Aerial Robotics: Exploring Accuracy-Computation Trade-off for Local Image Descriptors

“…Visual place recognition is critical for developing a practical and self-reliant UAV that does not require an external tracking or positioning system [16]. The fundamental task for VPR is building and searching an image database to determine if a previously visited location is detected.…”