In this work, we research and evaluate multiple pose-graph fusion strategies for vehicle localization. We focus on fusing a single absolute localization system, i.e. automotivegrade Global Navigation Satellite System (GNSS) at 1 Hertz, with a single relative localization system, i.e. vehicle odometry at 25 Hertz. Our evaluation is based on 180 Km long vehicle trajectories that are recorded in highway, urban and rural areas, and that are accompanied with post-processed Real Time Kinematic GNSS as ground truth. The results exhibit a significant reduction in the error’s standard deviation by 18% but the bias in the error is unchanged, when compared to non-fused GNSS. We show that the underlying principle is the fact that errors in GNSS readings are highly correlated in time. This causes a bias that cannot be compensated for by using the relative localization information from the odometry, but it can reduce the standard deviation of the error.

An Experimental Study on Relative and Absolute Pose Graph Fusion for Vehicle Localization

The Simultaneous Localization and Mapping (SLAM) technique has been used for autonomous navigation of mobile systems; now, its applications have been extended to 3D data acquisition of indoor environments. In order to reconstruct 3D scenes of indoor space, the kinematic 3D laser scanning system, developed herein, carries three laser range finders (LRFs): one is mounted horizontally for system-position correction and the other two are mounted vertically to collect 3D point-cloud data of the surrounding environment along the system’s trajectory. However, the kinematic laser scanning results can be impaired by errors resulting from sensor misalignment. In the present study, the bore-sight calibration of multiple LRF sensors was performed using a specially designed double-deck calibration facility, which is composed of two half-circle-shaped aluminum frames. Moreover, in order to automatically achieve point-to-point correspondences between a scan point and the target center, a V-shaped target was designed as well. The bore-sight calibration parameters were estimated by a constrained least squares method, which iteratively minimizes the weighted sum of squares of residuals while constraining some highly-correlated parameters. The calibration performance was analyzed by means of a correlation matrix. After calibration, the visual inspection of mapped data and residual calculation confirmed the effectiveness of the proposed calibration approach.

/pdf/bore-sight-calibration-of-multiple-laser-range-finders-for-1rbei2tbff.pdf

Bore-Sight Calibration of Multiple Laser Range Finders for Kinematic 3D Laser Scanning Systems.

The usage of a global positioning system (GPS) corrected inertial navigation system (INS) seems to be advantageous to camera-based pose estimation algorithms for outdoor navigation. The GPS signals lead to a geographical location with an accuracy of a few meters or even up to some centimeters in a setup utilizing correction data. Performing the pose estimation with a camera system has several disadvantages. The high amount of data has to be processed and the robustness may vary because of different environment and weather conditions. Nevertheless, cameras are the most used sensors for various tasks with unmanned aerial systems (UAS) because of their low cost and weight. The research results for vision-based simultaneous localization and mapping (SLAM) algorithms show the great capabilities of vision-based navigation. Therefore, the possibilities of cameras should be investigated for outdoor navigation in conjunction with an INS or as standalone backup solution in the case of an INS failure. The vision-based pose estimation using a SLAM algorithm provides considerable additional value by providing a map of the observed area. In contrast to keypoint-based approaches, the usage of a semi-dense formulation leads to 3D information for all image regions with gradients and results in a detailed 3D reconstruction. However, the main problem of the monocular SLAM remains the same. The visual pose estimations, and consequently the generated world, suffer from the unknown scale and the corresponding scale drift. In this work we develop a large scale semi-dense SLAM algorithm for the outdoor navigation of UAS equipped with a camera and an INS. This sensor combination increases the robustness of the visual pose estimation process and enables the generation of a very detailed, metric scaled map.

Scaling the world of monocular SLAM with INS-measurements for UAS navigation

A novel approach towards depth map super-resolution using multi-view uncalibrated photometric stereo is presented. Practically, an LED light source is attached to a commodity RGB-D sensor and is used to capture objects from multiple viewpoints with unknown motion. This non-static camera-to-object setup is described with a nonconvex variational approach such that no calibration on lighting or camera motion is required due to the formulation of an end-to-end joint optimization problem. Solving the proposed variational model results in high resolution depth, reflectance and camera pose estimates, as we show on challenging synthetic and real-world datasets.

/pdf/inferring-super-resolution-depth-from-a-moving-light-source-5fg5u45rgn.pdf

Inferring Super-Resolution Depth from a Moving Light-Source Enhanced RGB-D Sensor: A Variational Approach

In this paper we present an approach for performing system calibration of a sensor suite consisting of a fixed mounted camera and an inertial navigation system without the usage of a photogrammetric calibration site. The aim of the presented work is to obtain accurate direct georeferencing of camera images collected with manned or unmanned aerial systems for search and rescue tasks. These time-critical missions require a straightforward calibration which can be performed without additional equipment. This induces an in-flight calibration of the intrinsic camera parameters in addition to the mounting offsets between the camera and the inertial navigation system. The optimization of these values can be done by introducing them as parameters into a bundle adjustment process. We show how to solve this by exploiting a graph optimization framework. The evaluation of the proposed approach with data from flight experiments leads to an improvement of roughly factor six compared to a terrestrial calibration.

/pdf/ins-camera-calibration-without-ground-control-points-1ysrz3gnik.pdf

INS-camera calibration without ground control points

. In this paper, we present a graph based approach for performing the system calibration of a sensor suite containing a fixed mounted camera and an inertial navigation system. The aim of the presented work is to obtain accurate direct georeferencing of camera images collected with small unmanned aerial systems. Prerequisite for using the pose measurements from the inertial navigation system as exterior orientation for the camera is the knowledge of the static offsets between these devices. Furthermore, the intrinsic parameters of the camera obtained in a laboratory tend to deviate slightly from the values during flights. This induces an in-flight calibration of the intrinsic camera parameters in addition to the mounting offsets between the two devices. The optimization of these values can be done by introducing them as parameters into a bundle adjustment process. We show how to solve this by exploiting a graph optimization framework, which is designed for the least square optimization of general error functions.

/pdf/a-graph-based-bundle-adjustment-for-ins-camera-calibration-k6mepxi8zc.pdf

D. Bender

Papers

A graph based bundle adjustment for ins-camera calibration