Results 1 -
6 of
6
Self-Calibration and Visual SLAM with a Multi-Camera System on a Micro Aerial Vehicle
"... Abstract—The use of a multi-camera system enables a robot to obtain a surround view, and thus, maximize its perceptual awareness of its environment. An accurate calibration is a nec-essary prerequisite if vision-based simultaneous localization and mapping (vSLAM) is expected to provide reliable pose ..."
Abstract
-
Cited by 2 (2 self)
- Add to MetaCart
(Show Context)
Abstract—The use of a multi-camera system enables a robot to obtain a surround view, and thus, maximize its perceptual awareness of its environment. An accurate calibration is a nec-essary prerequisite if vision-based simultaneous localization and mapping (vSLAM) is expected to provide reliable pose estimates for a micro aerial vehicle (MAV) with a multi-camera system. On our MAV, we set up each camera pair in a stereo configuration. We propose a novel vSLAM-based self-calibration method for a multi-sensor system that includes multiple calibrated stereo cameras and an inertial measurement unit (IMU). Our self-calibration estimates the transform with metric scale between each camera and the IMU. Once the MAV is calibrated, the MAV is able to estimate its global pose via a multi-camera vSLAM implementation based on the generalized camera model. We propose a novel minimal and linear 3-point algorithm that uses inertial information to recover the relative motion of the MAV with metric scale. Our constant-time vSLAM implementation with loop closures runs on-board the MAV in real-time. To the best of our knowledge, no published work has demonstrated real-time on-board vSLAM with loop closures. We show experimental results in both indoor and outdoor environments. The code for both the self-calibration and vSLAM is available as a set of ROS packages at
Unsupervised Learning of Threshold for Geometric Verification in Visual-Based Loop-Closure
"... Abstract — A potential loop-closure image pair passes the geometric verification test if the number of inliers from the computation of the geometric constraint with RANSAC exceed a pre-defined threshold. The choice of the threshold is critical to the success of identifying the correct loop-closure i ..."
Abstract
-
Cited by 1 (0 self)
- Add to MetaCart
(Show Context)
Abstract — A potential loop-closure image pair passes the geometric verification test if the number of inliers from the computation of the geometric constraint with RANSAC exceed a pre-defined threshold. The choice of the threshold is critical to the success of identifying the correct loop-closure image pairs. However, the value for this threshold often varies for different datasets and is chosen empirically. In this paper, we propose an unsupervised method that learns the threshold for geometric verification directly from the observed inlier counts of all the potential loop-closure image pairs. We model the distributions of the inlier counts from all the potential loop-closure image pairs with a two components Log-Normal mixture model- one component represents the state of non loop-closure and the other represents the state of loop-closure, and learn the parameters with the Expectation-Maximization algorithm. The intersection of the Log-Normal mixture distributions is the optimal threshold for geometric verification, i.e. the threshold that gives the minimum false positive and negative loop-closures. Our algorithm degenerates when there are too few or no loop-closures and we propose the χ2 test to detect this degeneracy. We verify our proposed method with several large-scale datasets collected from both the multi-camera setup and stereo camera. I.
1 Using Photographs to Build and Augment 3D Models
"... Fig. 1: Part of an urban reconstruction computed via Structure-from-Motion and dense reconstruction obtained solely from images. The gray pyramids visualize camera location an orientation for captured photographs. This paper presents an overview over existing techniques in the field of computer visi ..."
Abstract
- Add to MetaCart
(Show Context)
Fig. 1: Part of an urban reconstruction computed via Structure-from-Motion and dense reconstruction obtained solely from images. The gray pyramids visualize camera location an orientation for captured photographs. This paper presents an overview over existing techniques in the field of computer vision for building digital 3D models and for augmenting them with additional photographs. In terms of 3D modelling we illustrate a fully automatic approach for the alignment of scans without the need for any artificial markers or manual interaction. In addition we show how to create entire models solely from images (cf. Fig. 1) up to the scale of whole cities. For the task of image location wrt. an existing model, we differentiate between urban, man-made environments and landscapes. We describe approaches for both cases and demonstrate how novel photographs can augment the 3D model in order to create a richer representation of an environment. We keep explanations at a higher level such that researchers from different fields are provided with a good overview; however, we reference numerous related works for the interested reader.
A Minimal Solution to the Rolling Shutter Pose Estimation Problem
"... Abstract — Artefacts that are present in images taken from a moving rolling shutter camera degrade the accuracy of absolute pose estimation. To alleviate this problem, we introduce an addition linear velocity in the camera projection matrix to approximate the motion of the rolling shutter camera. In ..."
Abstract
- Add to MetaCart
(Show Context)
Abstract — Artefacts that are present in images taken from a moving rolling shutter camera degrade the accuracy of absolute pose estimation. To alleviate this problem, we introduce an addition linear velocity in the camera projection matrix to approximate the motion of the rolling shutter camera. In partic-ular, we derive a minimal solution using the Gröbner Basis that solves for the absolute pose as well as the motion of a rolling shutter camera. We show that the minimal problem requires 5-point correspondences and gives up to 8 real solutions. We also show that our formulation can be extended to use more than 5-point correspondences. We use RANSAC to robustly get all the inliers. In the final step, we relax the linear velocity assumption and do a non-linear refinement on the full motion, i.e. linear and angular velocities, and pose of the rolling shutter camera with all the inliers. We verify the feasibility and accuracy of our algorithm with both simulated and real-world datasets. I.
Leveraging Image-Based Localization for Infrastructure-Based Calibration of a Multi-Camera Rig
"... Most existing calibration methods for multi-camera rigs are computationally expensive, use installations of known fiducial markers, and require expert supervision. We propose an alternative approach called infrastructure-based calibration that is efficient, requires no modification of the infrastruc ..."
Abstract
- Add to MetaCart
(Show Context)
Most existing calibration methods for multi-camera rigs are computationally expensive, use installations of known fiducial markers, and require expert supervision. We propose an alternative approach called infrastructure-based calibration that is efficient, requires no modification of the infrastructure (or calibration area), and is completely unsupervised. In infrastructure-based calibration, we use a map of a chosen calibration area and lever-age image-based localization to calibrate an arbitrary multi-camera rig in near real-time. Due to the use of a map, before we can apply infrastructure-based calibration, we have to run a survey phase once to generate a map of the calibration area. In this survey phase, we use a survey vehicle equipped with a multi-camera rig and a calibrated odom-etry system, and SLAM-based self-calibration to build the map which is based on nat-ural features. The use of the calibrated odometry system ensures that the metric scale of the map is accurate. Our infrastructure-based calibration method does not assume an overlapping field of view between any two cameras, and does not require an initial guess of any extrinsic parameter. Through extensive field tests on various ground vehi-cles in a variety of environments, we demonstrate the accuracy and repeatability of the infrastructure-based calibration method for calibration of a multi-camera rig. The code for our infrastructure-based calibration method is publicly available as part of the CamOdoCal library at
Autonomous Robots manuscript No. (will be inserted by the editor) Self-Calibration and Visual SLAM with a Multi-Camera System on a Micro Aerial Vehicle
"... Abstract The use of a multi-camera system enables a robot to obtain a surround view, and thus, maximize its perceptual awareness of its environment. If vision-based simultaneous localization and mapping (vSLAM) is expected to provide reliable pose estimates for a micro aerial vehicle (MAV) with a mu ..."
Abstract
- Add to MetaCart
(Show Context)
Abstract The use of a multi-camera system enables a robot to obtain a surround view, and thus, maximize its perceptual awareness of its environment. If vision-based simultaneous localization and mapping (vSLAM) is expected to provide reliable pose estimates for a micro aerial vehicle (MAV) with a multi-camera system, an accurate calibration of the multi-camera system is a necessary prerequisite. We propose a novel vSLAM-based self-calibration method for a multi-camera system that includes at least one calibrated stereo camera, and an arbitrary number of monocular cameras. We assume overlapping fields of view to only exist within stereo cameras. Our self-calibration estimates the inter-camera trans-forms with metric scale; metric scale is inferred from cal-ibrated stereo. On our MAV, we set up each camera pair in a stereo configuration which facilitates the estimation of the MAV’s pose with metric scale. Once the MAV is cal-ibrated, the MAV is able to estimate its global pose via a multi-camera vSLAM implementation based on the general-ized camera model. We propose a novel minimal and linear 3-point algorithm that uses relative rotation angle measure-ments from a 3-axis gyroscope to recover the relative mo-tion of the MAV with metric scale from 2D-2D feature cor-respondences. This relative motion estimation does not in-volve scene point triangulation. Our constant-time vSLAM
