This paper presents an Unscented Kalman Filter approach to visual-inertial odometry with sparse inequality map constraints. The system setup is motivated by a planetary swarm-exploration scenario, in ...which agile and light-weight agents (e.g. UAVs) navigate in an environment that was already partially mapped by complementary swarm vehicles (e.g. rovers). It is based on a state-of-the-art monocular system that facilitates state updates via three-view geometrical constraints of matched features. An IMU provides measurements for state prediction between successive updates. Our central contribution to the state-of the-art is the introduction of a method to incorporate inequality constraints on accessible space. These are given in the form of ordered sparse coordinates with associated vectors that identify accessible areas. We present experiments with data from the KITTI Vision Benchmark Suite, which contains all necessary data types acquired by an automotive system, along with simulated map constraints. Our results illustrate the potential and limitations of such sparse inequality constraints to correct for the drift of the odometry system.
The horizontal positioning accuracy of Global Navigation Satellite System receivers is in general two times higher than the vertical positioning accuracy. The integration of barometric height ...information improves in particular the vertical positioning accuracy. In this paper, we integrate differential air pressure measurements into Real-Time Kinematic (RTK) positioning with a Kalman filter. W e show that t he d ifferential a ir pressure measurements enable a faster convergence of the float RTK solution and a more reliable ambiguity fixing. T hereby, the proposed method is especially attractive for improving the RTK performance after temporary GNSS outages.
Precise point positioning with satellite navigation signals requires knowledge of satellite code and phase biases. In this paper, a new multi-stage method is proposed for estimating of these biases ...using measurements from a geodetic network. The method first subtracts all available a priori knowledge on orbits, satellite clocks and multipath from the measurements to reduce their dynamics. Secondly, satellite phase biases, ionospheric delays, carrier phase integer ambiguities and the geometry combining all non-dispersive parameters are jointly estimated in a Kalman filter. Finally, the a posteriori geometry estimates are refined in a second Kalman filter for the computation of orbital errors, code biases and tropospheric delays. As the first Kalman filter introduces time correlation, a generalized Kalman filter for colored measurement noise is applied in the second stage. The proposed algorithm is applied to dual frequency GPS measurements from a local geodetic network in Germany. A remarkable bias stability with variations of less than 3 cm over 4 hours is observed.
A precise position and attitude information is essential for autonomous driving of any vehicle. Low-cost GNSS receivers and antennas can provide a precise attitude and drift-free position ...information. However, severe code multipath, frequent half cycle slips and losses of lock might temporarily reduce the accuracy. Inertial sensors are robust to GNSS signal interruption and very precise over short time frames, which enables a reliable cycle slip correction. However, low-cost inertial sensors suffer from a substantial drift. In this paper, we propose a tightly coupled position and attitude determination method for two low-cost GNSS receivers, a gyroscope and an accelerometer. It improves classical tightly coupled solutions by including a synchronization correction, by the estimation of the code multipath for each satellite and receiver, and by the additional determination of satellite-satellite single difference ambiguities. The proposed method was verified in a test drive. We obtained a heading with an accuracy of 0.25°/baseline length m and an absolute position with an accuracy of 1 m.
Global Navigation Satellite Systems (GNSSs) may become a viable guidance means for safety-critical applications, such as the final approach and landing phases of a flight. To this purpose, ...Ground-Based Augmentation Systems (GBAS) are designed to enhance the navigation service, in terms of both accuracy and integrity. One of the tasks performed by GBAS stations is the timely and reliable detection of atmospheric disturbances that may compromise safety through biased or erroneous solutions. We address in this work the problem of detecting bi-dimensional ionospheric disturbances via GNSS carrier phase measurements from small-scale networks. Carrier phase measurements enable higher sensitivity for bias detection, but their inherent ambiguous nature has to be properly addressed. The reliable detection of biases is performed through standard Detection, Identification and Adaptation (DIA) techniques, and the impact of both observation noise and local baseline geometry (lengths and mutual orientations) is analyzed.
This study focuses on detecting invalid regimes of machine learning models through novelty detection algorithms. We apply them to a two-dimensional test case. Our results illustrate the impact of ...data noise and different hyperparameter settings.
Common field-of-view of cameras in robotic swarms Chen Zhu; Bamann, Christoph; Henkel, Patrick ...
2013 IEEE/RSJ International Conference on Intelligent Robots and Systems,
2013-Nov.
Conference Proceeding
Cooperative swarms of robots with cameras can provide stereo and multi-view vision. They are robust against failures and introduce diversity in the case of poor conditions. Cameras are used both for ...data acquisition and navigation. Additional sensors also play a role in the SLAM tasks (Simultaneously Localization and Mapping). The independent mobility of the robots/cameras leads to a situation in which the common field-of-view (FOV) of cameras changes continuously. The present paper addresses the task of acquiring and tracking the cameras FOVs. State-of-the-art FOV characterization techniques count feature points or do image segmentation. These methods are often not accurate or complex. We propose an adaptive common FOV detection method based on fuzzy plane clustering. The performance of the method is shown to be invariant under baseline scaling. An autonomous grouping algorithm is further proposed with respect to both distance of robots and overlapping FOV of cameras.
Recently, low-cost dual constellation receivers with simultaneous tracking of GPS and GLONASS satellites and raw carrier phase output have become available. This enables a faster and more reliable ...ambiguity fixing especially in areas with limited satellite visibility, e.g. street canyons. In this paper, we present a combined GPS/GLONASS RTK positioning with code multipath estimation for low-cost receivers. We use double difference measurements and take the integer property of both the GPS and GLONASS ambiguities into account. For GLONASS, this requires a reparametrization of the double difference ambiguities and a subsequent parameter mapping. The re-parameterized measurement models are used for ambiguity fixing with real measurements from two low-cost dual constellation GNSS receivers. We obtain residuals of a few centimeters for both GPS and GLONASS fixed carrier phases.