Improve Localization Accuracy Research Articles

An improved lightweight deep neural network with knowledge distillation for local feature extraction and visual localization using images and LiDAR point clouds

Visual localization nowadays is a research hotspot in computer vision and photogrammetry. It can provide meter level or higher localization accuracy under the conditions without GPS signals. However, achieving efficient, robust and high-accuracy visual localization under the condition of day-night changes is still challenging. To deal with this problem, we develop an improved lightweight deep neural network with knowledge distillation to efficiently extract deep local features from imagery while maintaining strong robustness for day-night visual localization. Furthermore, to further improve the accuracy of visual localization, we use aligned dense LiDAR point clouds and imagery collected by a new portable camera-LiDAR integrated device to build a prior map, and directly utilize the 2D-3D correspondences between 2D local feature points extracted by our lightweight network and 3D laser points retrieved from the prior map for localization. Moreover, we build our own ground truth point cloud dataset at 5 cm accuracy to evaluate the accuracy of the constructed prior map as well as a day-night dataset including prior map and verification data for the evaluation of the proposed visual localization method. The experimental results prove that our visual localization method achieves a balance between the efficiency and robustness while improving localization accuracy for day-night visual localization. In a comparison with a variety of state-of-the-art local feature extraction methods based on deep neural networks, our lightweight network has the least number of parameters (0.2 million) and reaches the highest feature extraction efficiency (92 frames per second), which is on par with that of the classic real-time ORB feature extraction method. Furthermore, our network remains competitive with other advanced deep local feature extraction networks in feature matching and day-night visual localization. In addition, evaluations performed on our own dataset demonstrate that our visual localization method using images and LiDAR point clouds provides a localization error of 1.2 m under the conditions of day-night changes, which is much smaller than those achieved by a state-of-the-art, purely visual localization method.

A Constrained Kalman Filter for Wi-Fi-Based Indoor Localization with Flexible Space Organization.

This paper presents a constrained Kalman filter for Wi-Fi-based indoor localization. The contribution of this work is to introduce constraints on the object speed and to provide a numerically optimized form for fast computation. The proposed approach is suitable to flexible space organization, as in warehouses, and when objects can be spun around, for example barcode readers in a hand. We experimented with the proposed technique using a robot and three devices, on five different journeys, in a 6000 m2 warehouse equipped with six Wi-Fi access points. The results highlight that the proposed approach provides a 19% improvement in localization accuracy.

Effects of Stimulation Position and Frequency Band on Auditory Spatial Perception with Bilateral Bone Conduction.

Virtual sound localization tests were conducted to examine the effects of stimulation position (mastoid, condyle, supra-auricular, temple, and bone-anchored hearing aid implant position) and frequency band (low frequency, high frequency, and broadband) on bone-conduction (BC) horizontal localization. Non-individualized head-related transfer functions were used to reproduce virtual sound through bilateral BC transducers. Subjective experiments showed that stimulation at the mastoid gave the best performance while the temple gave the worst performance in localization. Stimulation at the mastoid and condyle did not differ significantly from that using air-conduction (AC) headphones in localization accuracy. However, binaural reproduction at all BC stimulation positions led to similar levels of front-back confusion (FBC), which were also comparable to that with AC headphones. Binaural BC reproduction with high-frequency stimulation led to significantly higher localization accuracy than with low-frequency stimulation. When transcranial attenuation (TA) was measured, the attenuation became larger at the condyle and mastoid, and increased at high frequencies. The experiments imply that larger TAs may improve localization accuracy but do not improve FBC. The present study indicates that the BC stimulation at the mastoid and condyle can effectively convey spatial information, especially with high-frequency stimulation.

Iterative Implementation Method for Robust Target Localization in a Mixed Interference Environment

For the problem of target localization under the multipath propagation environment, the existing methods are mainly restricted to the limited prior information of complex reflections, especially when the target is embedded in a mixed interference environment. They may suffer from performance degradation due to the shortage of target classification ability. To address this problem, we propose a target localization method based on iterative implementation with semiunitary constraint and eigen-decomposition technique, where a practical propagation scenario based on the spherical Earth model is considered. Compared to the previous works, the proposed method can automatically distinguish a real target from the mixed interference environment with improved localization accuracy. Neither additional decorrelation preprocessing nor prior information of the dynamic scenario is required. Both simulations and real data experiments validate the effectiveness and robustness of the proposed method.

A Narrow Beam Steering Antenna Array for Indoor Positioning Systems Based on Wireless Sensor Network

This paper proposes the 2.45 <i>GHz</i> planar Beam Steering Antenna Array for Indoor Positioning Systems. The antenna array consists of a phase shifter based on a 4 &#x00D7; 4 Butler matrix, and an array of four dipole Yagi antennas. By choosing one of four beams using a switching controller, the main beam of the antenna can steer in four directions from +37&#x00B0;, &#x2013;12&#x00B0;, +12&#x00B0; to &#x2013;36&#x00B0;. The antenna has a narrow fan beam with a beamwidth in the azimuth plane of from +21.5&#x00B0; to +24.5&#x00B0; and in the elevation plane of approximately 90&#x00B0; which effectively minimizes multipath signals. These advantages are suitable for positioning systems to increase location accuracy. The antenna achieves a peak gain of from 9.1 to 9.8 <i>dBi</i> in four directions, a wide band of 400 <i>MHz</i>. The fully electronically steerable antenna prototype has been designed using CST software, simulated based on the finite element method, and fabricated on the RO4003C substrate as well as measured. The antenna is implemented in the Indoor Positioning System using three different location methods, including trilateration, triangulation, and fingerprinting, to highlight the antenna&#x2019;s advantages in improving location accuracy with the ratio between the mean location error and area of 0.051, 0.033, and 0.023 respectively.

A Novel Low-Light Catenary Image Enhancement Approach for CSCs Detection in High-Speed Railways

The catenary system is essential for ensuring the stable energy transmission of trains in high-speed railways. The non-contact catenary detection is a promising monitoring method, where the catenary image is captured by an industrial camera mounted on the inspection vehicle. The image quality is susceptible to the limitations of the environment and equipment, which adversely affects the catenary location and detection accuracy. In this paper, we propose an unsupervised learning-based catenary image enhancement method for improving localization accuracy. First, the enhancement model is optimized to enhance the catenary image quality, making it sharper and more conducive to detection. Subsequently, an advanced small target location approach, called TPH-YOLOv5, is used to locate the catenary components. Finally, we compared the localization performance of the enhanced image with the low-light image. The experiment results show that the proposed method can effectively enhance the quality of low-light catenary images and improve the positioning accuracy.

Robust GPS-Vision Localization via Integrity-Driven Landmark Attention

<h3>Abstract</h3> For robust GPS-vision navigation in urban areas, we propose an integrity-driven landmark attention (ILA) technique via stochastic reachability. Inspired by cognitive attention in humans, we perform convex optimization to select a subset of landmarks from GPS and vision measurements that maximizes integrity-driven performance. Given known measurement error bounds in non-faulty conditions, our ILA technique follows a unified approach to address both GPS and vision faults and is compatible with any off-the-shelf estimator. We analyze measurement deviation to estimate the stochastic reachable set of positions associated with each landmark, which is parameterized via probabilistic zonotope (p-zonotope). We apply set union to formulate a p-zonotopic cost that represents the size of position bounds based on landmark inclusion/exclusion. We jointly minimize the p-zonotopic cost and maximize the number of landmarks via convex relaxation. For an urban data set, we demonstrate improved localization accuracy and robust predicted availability for a pre-defined risk and alert limit.

Device-Free Localization Using Enhanced Channel Selection and a Distance-Based Elliptical Model

Device-free localization (DFL) based on wireless sensor networks (WSNs) is a technology that can detect and locate a person by measuring the changes in received signals without the need for any wireless devices. As an emerging important technology in WSNs, radio tomographic imaging (RTI) has received increasing attention. However, there is much room to improve localization accuracy in RTI. To address this issue, an enhanced channel-selection method and a new distance-based elliptical model are proposed to improve the localization accuracy. The enhanced frequency channel-selection method selects two channels with the lowest received signal strength (RSS) variances to collect data. This approach is more robust to environmental change. The new distance-based elliptical model is based on the distance between the voxels and sensors. Meanwhile, the communication links are divided into line-of-sight (LOS) paths and nonline-of-sight (NLOS) paths. Experimental results demonstrate that the proposed algorithm improves the accuracy of positioning by up to 44.8% over some state-of-the-art RTI methods with low cost.

Harris hawks optimization algorithm and BP neural network for ultra-wideband indoor positioning.

Traditional back propagation neural networks (BPNNs) for ultrawideband (UWB) indoor localization can effectively improve localization accuracy, although there is high likelihood of becoming trapped in nearby minima. To solve this problem, the random weights and thresholds of the BPNN are optimized using the Harris Hawks optimization algorithm (HHO) to obtain the optimal global solution to enhance the UWB indoor positioning accuracy and NLOS resistance. The results show that the predicted trajectory of the HHO and BPNN hybrid algorithm (HHO-BP) matches the actual position in the two-dimensional localization scenario with four base stations; the optimized average positioning error is effectively reduced in both indoor LOS and NLOS environments. In the LOS environment, the total mean error of the traditional BPNN algorithm is 6.52 cm, which is 26.99% better than the UWB measurement error; in the NLOS environment, the total mean error of the conventional BPNN is 14.82 cm, which is 50.08% better than the UWB measurement error. The HHO-BP algorithm is further optimized on this basis, and the total mean error in the LOS environment is 4.50 cm, which is 22.57% better than the conventional BPNN algorithm; in the NLOS environment, the total mean error is 9.56 cm, which is 17.54% better than the conventional BPNN algorithm. The experimental findings suggest that the approach has greater calibration accuracy and stability than BPNN, making it a viable choice for scenarios requiring high positional precision.

An infrasound source localisation algorithm for improving location accuracy of gas pipeline leakage detection system

An infrasound source localisation algorithm for improving location accuracy of gas pipeline leakage detection system

An infrasound source localisation algorithm for improving location accuracy of gas pipeline leakage detection system

An infrasound source localisation algorithm for improving location accuracy of gas pipeline leakage detection system

Dynamic Scene’s Laser Localization by NeuroIV-Based Moving Objects Detection and LiDAR Points Evaluation

Accurate localization is an important component for the vehicle&#x2019;s autonomous navigation. The appearance of the moving objects may lead to feature-matching error with the map features, thereby causing a serious decline of localization accuracy. Neuromorphic vision sensor (NeuroIV) is a kind of dynamic vision sensor, with the properties of high temporal resolution, movement capture, and lightweight computation. In view of this, this research proposes to combine the NeuroIV and LIDAR points to acquire the static landmark features and robust navigation localization. However, as a younger and smaller research field compared to RGB computer vision, NeuroIV vision is rarely associated with the intelligent vehicle. For this purpose, we built a novel dataset recorded by NeuroIV sensor, and a state-of-the-art YOLO-small network is designed to detect the moving objects with the dataset. In order to completely deduct the whole dynamic zones, a sensors&#x2019; novel fusion model is built by the zones&#x2019; segmentation and matching, so the LIDAR&#x2019;s static environment is obtained completely by the remained points. By evaluating different types of LIDAR points, the feature-matching error can be alleviated further, making the localization is more accurate. Together with qualitative and quantitative results, this work provides a moving objects&#x2019; detection improvement of 14.13 % mAP with the new NeuroIV dataset, and an obvious localization accuracy improvement with LIDAR points&#x2019; evaluation.

OGI-SLAM2: A Hybrid Map SLAM Framework Grounded in Inertial-Based SLAM

Indoor localization using mobile sensors has gained momentum lately. However, indoor pedestrian localization with low-cost IMU has not been practically applied, and the main challenge is that it is difficult for low-cost IMU to obtain high-precision localization performance over a long period of time. In this paper, we propose a multi-source fusion indoor pedestrian autonomous localization framework combining landmark maps and occupancy grid maps. Combining the ideas of optimization and occupancy grid map-based inertial SLAM algorithm(OGI-SLAM), we maintain the short- and medium-term localization accuracy with OGI-SLAM, and achieve long-term localization accuracy improvement with keypoint maps. Without prior knowledge of the map, we use user motion features and environmental features to extract keypoints in the trajectory and construct keypoint descriptors. Combining the uncertainty of system location and the similarity of environmental features, we propose a keypoint extraction and association method to solve the problem of inaccurate particle filtering loop closing with low-cost IMU. In addition, a two-step trajectory optimization method is proposed, which uses bundle adjustment to correct the positioning error of keypoints and models the error correction of common points between keypoints as affine transformation, avoiding the large matrix inversion operation in the optimization process. Simulations and field trial evaluations show that the system is able to achieve a maximum positioning accuracy of 1.64 meters with low-performance IMU on board, representing a 57% maximum improvement in positioning accuracy compared to OGI-SLAM. This highlights the promise of OGI-SLAM2 as an unconventional indoor positioning method.

Localization of isotropic and anisotropic wireless sensor networks in 2D and 3D fields

Internet of Things (IoT) has changed the way people live by transforming everything into smart systems. Wireless Sensor Network (WSN) forms an important part of IoT. This is a network of sensor nodes that is used in a vast range of applications. WSN is formed by the random deployment of sensor nodes in various fields of interest. The practical fields of deployment can be 2D or 3D, isotropic or anisotropic depending on the application. The localization algorithms must provide accurate localization irrespective of the type of field. In this paper, we have reported a localization algorithm called Range Reduction Based Localization (RRBL). This algorithm utilizes the properties of hop-based and centroid methods to improve the localization accuracy in various types of fields. In this algorithm, the location unknown nodes identify the close-by neighboring nodes within a predefined threshold and localize themselves by identifying and reducing the probable range of existence from these neighboring nodes. The nodes which do not have enough neighbors are localized using the least squares method. The algorithm is tested in various irregular and heterogeneous conditions. The results are compared with a few state-of-the-art hop-based and centroid-based localization techniques. RRBL has shown an improvement in localization accuracy of 28% at 10% reference node ratio and 26% at 20% reference node ratio when compared with other localization algorithms.

A Survey of Recent Indoor Localization Scenarios and Methodologies.

Recently, various novel scenarios have been studied for indoor localization. The trilateration is known as a classic theoretical model of geometric-based indoor localization, with uniform RSSI data that can be transferred directly into distance ranges. Then, a trilateration solution can be algebraically acquired from theses ranges, in order to fix user’s actual location. However, the collected RSSI or other measurement data should be further processed and classified to lower the localization error rate, instead of using the raw data influenced by multi-path effect, multiple nonlinear interference and noises. In this survey, a large number of existing techniques are presented for different indoor network structures and channel conditions, divided as LOS (light-of-sight) and NLOS (non light-of-sight). Besides, the input measurement data such as RSSI (received signal strength indication), TDOA (time difference of arrival), DOA (distance of arrival), and RTT (round trip time) are studied towards different application scenarios. The key localization techniques like RSSI-based fingerprinting technique are presented using supervised machine learning methods, namely SVM (support vector machine), KNN (K nearest neighbors) and NN (neural network) methods, especially in an offline training phase. Other unsupervised methods as isolation forest, k-means, and expectation maximization methods are utilized to further improve the localization accuracy in online testing phase. For Bayesian filtering methods, apart from the basic linear Kalman filter (LKF) methods, nonlinear stochastic filters such as extended KF, cubature KF, unscented KF and particle filters are introduced. These nonlinear methods are more suitable for dynamic localization models. In addition to the localization accuracy, the other important performance features and evaluation aspects are presented in our paper: scalability, stability, reliability, and the complexity of proposed algorithms is compared in this survey. Our paper provides a comprehensive perspective to compare the existing techniques and related practical localization models, with the aim of improving localization accuracy and reducing the complexity of the system.

Know Where You are: A Practical Privacy-Preserving Semi-Supervised Indoor Positioning via Edge-Crowdsensing

In recent years, with the popularity of smartphones, the indoor positioning systems based on mobile crowdsensing (MCS) have gained considerable interest and exploit. However, it is still challenging to construct a largescale indoor positioning system. 1) In indoor positioning model, storage and computing resources are very important. 2) The calibration operation of data label and selection of model parameters require the operation of professionals. 3) User location privacy may be compromise, which greatly affects participant safety and enthusiasm. To solve these problems, our model firstly provides an edge-crowdsourcing indoor localization architecture to improve storage, computing power and response speed. Then, based on manifold regularization, a semi-supervised indoor localization model is determined by an adaptive manner in terms of both similarity and manifold structure, which reduces the workload of the positioning model and improve localization accuracy. In addition, we propose a new privacy-aware indoor localization algorithm based on secure multi-party computation to protect location privacy. Experimental results on real-world datasets show that, compared with the previous methods, our method improves accuracy by 0.87m, and in terms of time overhead of privacy protection, our method reduces the running time of the thousand seconds level.

SOMA: Subject-, object-, and modality-adapted precision atlas approach for automatic anatomy recognition and delineation in medical images.

In the multi-atlas segmentation (MAS) method, a large enough atlas set, which can cover the complete spectrum of the whole population pattern of the target object will benefit the segmentation quality. However, the difficulty in obtaining and generating such a large set of atlases and the computational burden required in the segmentation procedure make this approach impractical. In this paper, we propose a method called SOMA to select subject-, object-, and modality-adapted precision atlases for automatic anatomy recognition in medical images with pathology, following the idea that different regions of the target object in a novel image can be recognized by different atlases with regionally best similarity, so that effective atlases have no need to be globally similar to the target subject and also have no need to be overall similar to the target object. The SOMA method consists of three main components: atlas building, object recognition, and object delineation. Considering the computational complexity, we utilize an all-to-template strategy to align all images to the same image space belonging to the root image determined by the minimum spanning tree (MST) strategy among a subset of radiologically near-normal images. The object recognition process is composed of two stages: rough recognition and refined recognition. In rough recognition, subimage matching is conducted between the test image and each image of the whole atlas set, and only the atlas corresponding to the best-matched subimage contributes to the recognition map regionally. The frequency of best match for each atlas is recorded by a counter, and the atlases with the highest frequencies are selected as the precision atlases. In refined recognition, only the precision atlases are examined, and the subimage matching is conducted in a nonlocal manner of searching to further increase the accuracy of boundary matching. Delineation is based on a U-net-based deep learning network, where the original gray scale image together with the fuzzy map from refined recognition compose a two-channel input to the network, and the output is a segmentation map of the target object. Experiments are conducted on computed tomography (CT) images with different qualities in two body regions - head and neck (H&N) and thorax, from 298 subjects with nine objects and 241 subjects with six objects, respectively. Most objects achieve a localization error within two voxels after refined recognition, with marked improvement in localization accuracy from rough to refined recognition of 0.6-3mm in H&N and 0.8-4.9mm in thorax, and also in delineation accuracy (Dice coefficient) from refined recognition to delineation of 0.01-0.11 in H&N and 0.01-0.18 in thorax. The SOMA method shows high accuracy and robustness in anatomy recognition and delineation. The improvements from rough to refined recognition and further to delineation, as well as immunity of recognition accuracy to varying image and object qualities, demonstrate the core principles of SOMA where segmentation accuracy increases with precision atlases and gradually refined object matching.

Genetic Algorithm for Path Loss Model Selection in Signal Strength-Based Indoor Localization

Ranging methods using received signal strength (RSS) information are widely used for indoor localization because they can be easily implemented without conducting site surveys. However, range estimation with a single pathloss model produces considerable errors, which degrade the positioning performance. This problem mainly arises because the single pathloss model cannot reflect diverse indoor radio wave propagation characteristics. In this study, we develop a new overlapping multi-state model to consider multiple candidates of pathloss models including line-of-sight (LOS) and non-line-of-sight (NLOS) states, and propose an efficient way to select a proper model for each reference node involved in the localization process. To this end, we formulate a cost function whose value varies widely depending on the choice of pathloss model of each access point. Because the computational complexity to find an optimal channel model for each reference node exponentially increases with the number of reference nodes, we apply a genetic algorithm to significantly reduce the complexity so that the proposed method can be executed in real-time. Experimental validations with ray-tracing simulations and RSS measurements at a real site confirm the improvement of localization accuracy for Wi-Fi in indoor environments. The proposed method achieves up to 1.92 m mean positioning error under a practical indoor environment and produces a performance improvement of 31.09% over the benchmark scenario.

The multi-level classification and regression network for visual tracking via residual channel attention

Siamese networks used for target tracking has attracted widespread attention due to its balanced tracking accuracy and efficient execution speed. However, when there exist similar semantic information in the search area, it is difficult for most Siamese trackers to adapt to the interference of similar semantic information to target localization, which greatly affects the robustness of Siamese trackers. In order to effectively mine feature information and improve localization accuracy, this work proposes one Siamese multi-level classification and regression (SiamMCAR) framework. SiamMCAR first introduces the residual channel attention module into the template branch of Siamese subnetwork. By utilizing the relationship between feature channels to determine the channel weight of the target template feature, making the attention of extracting the template feature is focused on the channel feature of target foreground. Then, one multi-level classification and regression subnetwork containing multiple classification and regression modules is constructed. Feature maps of the output of different classification and regression modules are weighted and fused by using multiple trained weights, which enables the multi-level classification and regression subnetwork to obtain more results of classification and regression of the shallow cross-correlation response map, thereby making the localization more accurate. Extensive experiments on many benchmarks like OTB-50, OTB-100, TC-128, GOT-10K and LaSOT have proved that our SiamMCAR achieves excellent performance and runs at 28 FPS in various challenging tasks of target tracking.

PLF-VINS: Real-Time Monocular Visual-Inertial SLAM With Point-Line Fusion and Parallel-Line Fusion

This letter presents a real-time monocular visual-inertial simultaneous localization and mapping with point-line fusion and parallel-line fusion. The corner and line features provide plenty of information about object structures. In the 2D image plane, such corner and line features have a positional similarity. The corner feature represents an endpoint of an object's edge, and the line feature represents a straight edge. We propose two novel methods for fusing corner and line features to improve localization accuracy. The first method is for fusing corner and line features. Using the positional similarity of corner and line features, we search the relationship between two features by utilizing the proposed point-line coupled residual. The second method is for fusing parallel 3D lines. First, we search for line features clustered based on a vanishing point. Then, the outliers in the parallel 3D lines are removed using the proposed consistency check during the multi-view line clustering. Finally, the parallel 3D lines are used in the proposed parallel 3D line residual. Experimental results show that real-time localization accuracy is improved when two proposed residuals are integrated with the sliding-window optimization. The proposed PLF-VINS is compared with other state-of-the-art algorithms using the public EuRoC dataset.