Search

Light Detection and Ranging (LiDAR) based Simultaneous Localization and Mapping (SLAM) has drawn increasing interests in autonomous driving. However, LiDAR-SLAM suffers from accumulating errors which can be significantly mitigated by Global Navigation Satellite System (GNSS). Precise Point Positioning (PPP), an accurate GNSS operation mode independent of base stations, gains growing popularity in unmanned systems. Considering the features of the two technologies, LiDAR-SLAM and PPP, this paper proposes a SLAM system, namely P3-LOAM (PPP based LiDAR Odometry and Mapping) which couples LiDAR-SLAM and PPP. For better integration, we derive LiDAR-SLAM positioning covariance by using Singular Value Decomposition (SVD) Jacobian model, since SVD provides an explicit analytic solution of Iterative Closest Point (ICP), which is a key issue in LiDAR-SLAM. A novel method is then proposed to evaluate the estimated LiDAR-SLAM covariance. In addition, to increase the reliability of GNSS in urban canyon environment, we develop a LiDAR-SLAM assisted GNSS Receiver Autonomous Integrity Monitoring (RAIM) algorithm. Finally, we validate P3-LOAM with UrbanNav, a challenging public dataset in urban canyon environment. Comprehensive test results prove that, in terms of accuracy and availability, P3-LOAM outperforms benchmarks such as Single Point Positioning (SPP), PPP, LeGO-LOAM, SPP-LOAM, and the loosely coupled navigation system proposed by the publisher of UrbanNav.

The problem of visual odometry (VO) and localization in extreme illumination conditions is widely concerned. In this paper, we propose a novel SLAM algorithm namely DVT-SLAM (Deep-learning based Visible-Thermal SLAM). It focuses on the fusion of thermal infrared image and visible image which have complementary advantages in characteristics. With the contrastive learning and the measurement of mutual information between multi-modal images, the first part of DVT-SLAM is the DVT-GAN network to fuse visible-thermal images and generate pseudo visible images at night. Given the generated images, visual odometry is applied for pose estimation base. Extensive evaluations are performed on the Brno Urban Dataset, a multi-modal dataset containing different time and weather conditions in diverse scenarios. Series of experiments show that DVT-SLAM is a robustness and suitability solution for single visible camera failures, which can reduce positioning error by half and achieve superior SLAM performance.

In this paper, a zero attracted Lorentzian (ZAL) algorithm is created for system identification applications. The proposed ZAL algorithm is realized by using l 1 -norm penalty on the Lorentzian-based cost function to exploit the sparseness of the existing nature signals. Moreover, a reweighting method is adopt to enhance the ability of the proposed ZAL algorithm, and the new algorithm is called as reweighting zero attracted Lorentzian (RZAL) algorithm. The proposed algorithms are presented briefly and investigated via computer simulations. The gotten results from the simulation demonstrate that the proposed algorithms are superior to the popular adaptive filtering algorithms under alpha-stable noise interference.

In this paper, we propose a novel method to detect rectangular and triangular traffic signs in natural scenes according to the geometric features of the rectangle and triangle traffic signs. It is a Gradient-direction-based method that mainly includes four stages: gradient information classification, filter templates-based line segments with different orientation extraction, structural corners detection and rectangle/triangle detection. The extracted corners are categorized into different sets according to the orientation of the line segments, by which we can quickly extract rectangle and triangle traffic signs in image. And the experimental results demonstrate under different conditions are shown that the proposed method can detect rectangular and triangular traffic signs in natural scenes efficiently and robustness.

The naturalness of the panoramic image is an important criterion for evaluating the result of image stitching. When the overlapping region of the image has a low-texture area, the alignment is hard to accomplish due to incorrect and inadequate feature point matching pairs, and the salient structures are easily destroyed due to the unreasonable position of the optimal seam. To solve the problem, we use line segment matching information to assist the alignment of low-texture regions and use superpixel segmentation regions to filter the unreliable point correspondences. An improved seam-finding algorithm is finally used to blend images together. The experimental results demonstrate that the proposed method can get more natural stitching results.

Precise relative navigation between unmanned aerial vehicles (UAVs) in a GNSS-denied environment is a critical technology enabling collaborative missions. Range-based relative navigation allows other onboard optical sensors to be used for missions instead of for tracking other formation members. However, range-based relative navigation suffers from a well-known ambiguous solution problem. Approaches developed in previous studies include the use of one or more anchors, optical-flow measuring velocity, or multiple range sensors to solve the observability of the estimated states. In this study, a novel range-based collaborative relative navigation algorithm for multiple fixed-wing UAVs is proposed by integrated ranging-sensor with other commonly used onboard sensors such as low-cost IMUs, where anchor, optical-flow measurement or multi ranging sensor is not required. A relative motion estimation model is established in a floating horizontal frame; the sensor biases are augmented in the model to solve the colored noise problem. The observability of the estimated states is analyzed, and the observable criteria to guide the flight are obtained by introducing the Lie derivative criteria. The relative position vector must not be a constant in both direction and distance, and non-rectilinear trajectory is also required to obtain observability. A decentralized estimation strategy based on a consensus extended Kalman filter is designed for the system, with several physical constraints on the estimation in constructing the consensus to improve the observability. The proposed algorithm is tested and verified using standard Monte Carlo simulations. The simulation results indicate that the full state is observable if the observable criteria are satisfied. The sensitivity of the relative localization accuracy to the sensors and motion is presented and discussed. It is shown that the uncertainties from ranging-sensors and IMUs contribute more to the performance of the estimation. The estimated uncertainties of the sensor biases are also promising, and can potentially be used to improve the absolute navigation of UAVs.

Aiming at the problem of low accuracy and effectiveness of complex system support capability assessment, an index system was constructed on the basis of studying the characteristics of support capability generation elements, in which the systematicness, comprehensiveness and the importance of key links of complex system support were considered, a complex system supportability evaluation model was proposed based on weight scores and key levels, and the score and level of the complex system supportability evaluation were obtained through comprehensive evaluation. A corresponding evaluation system was designed which provided the means and support for the informatization of support capability evaluation of complex systems.

In order to achieve high precision of initial alignment and inertial sensor calibration, this paper proposes a high accuracy multiplex two-position alignment method based on SINS (Strapdown Inertial Navigation System) with the aid of star sensor. Firstly, the Mars rover's approximate initial position is determined by the lander's landing position, and its initial attitude is obtained with the time and the output of star sensor. Based on that, a two-position alignment method aided by star sensor's precise attitude information is used to improve accuracy of initial alignment and inertial sensor calibration. It can take full advantages of both inertial navigation and celestial navigation. Finally, we deduce the relationships between estimation errors of accelerometer bias, estimation errors of misalignment angles and position errors. The rover's higher accuracy initial alignment is achieved by using the deduced relationships and the estimates of accelerometer bias to correct the rover's initial position and attitude. The simulation results demonstrate that the presented method can meet the high accuracy requirements of the rover's initial alignment and inertial sensor calibration.

Autonomous navigation is an important function for a Mars rover to fulfill missions successfully. It is a critical technique to overcome the limitations of ground tracking and control traditionally used. This paper proposes an innovative method based on SINS (Strapdown Inertial Navigation System) with the aid of star sensors to accurately determine the rover׳s position and attitude. This method consists of two parts: the initial alignment and navigation. The alignment consists of a coarse position and attitude initial alignment approach and fine initial alignment approach. The coarse one is used to determine approximate position and attitude for the rover. This is followed by fine alignment to tune the approximate solution to accurate one. Upon the completion of initial alignment, the system can be used to provide real-time navigation solutions for the rover. An autonomous navigation algorithm is proposed to estimate and compensate the accumulated errors of SINS in real time. High accuracy attitude information from star sensor is used to correct errors in SINS. Simulation results demonstrate that the proposed methods can achieve a high precision autonomous navigation for Mars rovers.