Search Results (1,708)

Three-dimensional (3D) object detection is crucial for autonomous driving, yet current PointPillar feature-based methods face challenges like under-segmentation, overlapping, and false detection, particularly in occluded scenarios. This paper presents a novel dual-stage improved PointPillar feature-based 3D object detection method (S2*-ODM) specifically designed to address these issues. The first innovation is the introduction of a dual-stage pillar feature encoding (S2-PFE) module, which effectively integrates both inter-pillar and intra-pillar relational features. This enhancement significantly improves the recognition of local structures and global distributions, enabling better differentiation of objects in occluded or overlapping environments. As a result, it reduces problems such as under-segmentation and false positives. The second key improvement is the incorporation of an attention mechanism within the backbone network, which refines feature extraction by emphasizing critical features in pseudo-images and suppressing irrelevant ones. This mechanism strengthens the network’s ability to focus on essential object details. Experimental results on the KITTI dataset show that the proposed method outperforms the baseline, achieving notable improvements in detection accuracy, with average precision for 3D detection of cars, pedestrians, and cyclists increasing by 1.04%, 2.17%, and 3.72%, respectively. These innovations make S2*-ODM a significant advancement in enhancing the accuracy and reliability of 3D object detection for autonomous driving. Full article

Tropospheric aerosols play an important role in the notable warming phenomenon and climate change occurring in the Arctic. The accuracy of Cloud–Aerosol Lidar with Orthogonal Polarization (CALIOP) aerosol optical depth (AOD) and the distribution of Arctic AOD based on the CALIOP Level 2 aerosol products and the Aerosol Robotic Network (AERONET) AOD data during 2006–2021 were analyzed. The distributions, trends, and three-dimensional (3D) structures of the frequency of occurrences (FoOs) of different aerosol subtypes during 2006–2021 are also discussed. We found that the CALIOP AOD exhibited a high level of agreement with AERONET AOD, with a correlation coefficient of approximately 0.67 and an RMSE of less than 0.1. However, CALIOP usually underestimated AOD over the Arctic, especially in wet conditions during the late spring and early summer. Moreover, the Arctic AOD was typically higher in winter than in autumn, summer, and spring. Specifically, polluted dust (PD), dust, and clean marine (CM) were the dominant aerosol types in spring, autumn, and winter, while in summer, ES (elevated smoke) from frequent wildfires reached the highest FoOs. There were increasing trends in the FoOs of CM and dust, with decreasing trends in the FoOs of PD, PC (polluted continental), and DM (dusty marine) due to Arctic amplification. In general, the vertical distribution patterns of different aerosol types showed little seasonal variation, but their horizontal distribution patterns at various altitudes varied by season. Furthermore, locally sourced aerosols such as dust in Greenland, PD in eastern Siberia, and ES in middle Siberia can spread to surrounding areas and accumulate further north, affecting a broader region in the Arctic. Full article

Quantum illumination (QI) is an entanglement-based protocol for improving LiDAR/radar detection of unresolved targets beyond what a classical LiDAR/radar of the same average transmitted energy can do. Originally proposed by Seth Lloyd as a discrete-variable quantum LiDAR, it was soon shown that his proposal offered no quantum advantage over its best classical competitor. Continuous-variable, specifically Gaussian-state, QI has been shown to offer a true quantum advantage, both in theory and in table-top experiments. Moreover, despite its considerable drawbacks, the microwave version of Gaussian-state QI continues to attract research attention. A recent QI study by Armanpreet Pannu, Amr Helmy, and Hesham El Gamal (PHE), however, has: (i) combined the entangled state from Lloyd’s QI with the channel models from Gaussian-state QI; (ii) proposed a new positive operator-valued measurement for that composite setup; and (iii) claimed that, unlike Gaussian-state QI, PHE QI achieves the Nair–Gu lower bound on QI target-detection error probability at all noise brightnesses. PHE’s analysis was asymptotic, i.e., it presumed infinite-dimensional entanglement. The current paper works out the finite-dimensional performance of PHE QI. It shows that there is a threshold value for the entangled-state dimensionality below which there is no quantum advantage, and above which the Nair–Gu bound is approached asymptotically. Moreover, with both systems operating with error-probability exponents 1 dB lower than the Nair–Gu bound, PHE QI requires enormously higher entangled-state dimensionality than does Gaussian-state QI to achieve useful error probabilities in both high-brightness (100 photons/mode) and moderate-brightness (1 photon/mode) noise. Furthermore, neither system has an appreciable quantum advantage in low-brightness (much less than 1 photon/mode) noise. Full article

The paper presents a fast method for 3D point cloud target extraction, addressing the challenge of time-consuming processing in LiDAR-based 3D point cloud data. The method begins with the acquisition of environmental 3D point cloud data using LiDAR, which is then projected onto a 2D cylindrical map. We propose a method for rapid target extraction from LiDAR-based 3D point cloud data, which includes key steps such as projection into 2D space, image processing for segmentation, and target extraction. A mapping matrix between the 2D grayscale image and the cylindrical projection is derived through Gaussian elimination. A target backtracking search algorithm is used to map the extracted target region back to the original 3D point cloud, enabling precise extraction of the 3D target points. Near-field experiments using hybrid solid-state LiDAR demonstrate the method’s effectiveness, requiring only 0.53 s to extract 3D target point clouds from datasets containing hundreds of thousands of points. Further, far-field rocket launch experiments show that the method can extract target point clouds within 158 milliseconds, with measured positional offsets of 0.2159 m and 0.1911 m as the rocket moves away from the launch tower. Full article

To mitigate the impact on detection performance caused by insufficient input information in 3D object detection based on single LiDAR data, this study designs three innovative modules based on the PointRCNN framework. Firstly, addressing the issue of the Multi-Layer Perceptron (MLP) in PointNet++ failing to effectively capture local features during the feature extraction phase, we propose the Adaptive Multilayer Perceptron (AMLP). Secondly, to prevent the problem of gradient vanishing due to the increased parameter scale and computational complexity of AMLP, we introduce the Channel Aware Residual module (CA-Res) in the feature extraction layer. Finally, in the head layer of the subsequent processing stage, we propose the Dynamic Attention Head (DA-Head) to enhance the representation of key features in the process of target detection. A series of experiments conducted on the KITTI validation set demonstrate that in complex scenarios, for the small target “Pedestrian”, our model achieves performance improvements of 2.08% and 3.46%, respectively, at the “Medium” and “Difficult” detection difficulty levels. To further validate the generalization capability of the Enhanced DetNet network, we deploy the trained model on the KITTI server and conduct a comprehensive evaluation of detection performance for the “Car”, “Pedestrian”, and “Cyclist” categories. Full article

Light detection and ranging (Lidar) is a key enabling technology for autonomous vehicles and drones. Its emerging implementations are based on photonic integrated circuits (PICs) and optical phased arrays (OPAs). In this work, we introduce a novel approach to the design of OPA Lidar antennas based on Si₃N₄ grating couplers. The well-established TriPleX platform and the asymmetric double stripe waveguide geometry with full etching are employed, ensuring low complexity and simple fabrication combined with the low-loss advantages of the platform. The design study aims to optimize the performance of the grating coupler-based radiators as well as the OPA, thus enhancing the overall capabilities of Si₃N₄-based Lidar. Uniform and non-uniform grating structures are considered, achieving θ and φ angle divergences of 0.9° and 32° and 0.54° and 25.41°, respectively. Also, wavelength sensitivity of 7°/100 nm is achieved. Lastly, the fundamental OPA parameters are investigated, and 35 dBi of peak directivity is achieved for an eight-element OPA. Full article

This study aimed to develop a real-time localization system for an AMR (autonomous mobile robot), which utilizes the Robot Operating System (ROS) Noetic version in the Ubuntu 20.04 operating system. RTAB-MAP (Real-Time Appearance-Based Mapping) is employed for localization, integrating with an RGB-D camera and a 2D LiDAR for real-time localization and mapping. The navigation was performed using the A* algorithm for global path planning, combined with the Dynamic Window Approach (DWA) for local path planning. It enables the AMR to receive velocity control commands and complete the navigation task. RTAB-MAP is a graph-based visual SLAM method that combines closed-loop detection and the graph optimization algorithm. The maps built using these three methods were evaluated with RTAB-MAP localization and AMCL (Adaptive Monte Carlo Localization) in a high-similarity long corridor environment. For RTAB-MAP and AMCL methods, three map optimization methods, i.e., TORO (Tree-based Network Optimizer), g2o (General Graph Optimization), and GTSAM (Georgia Tech Smoothing and Mapping), were used for the graph optimization of the RTAB-MAP and AMCL methods. Finally, the TORO, g2o, and GTSAM methods were compared to test the accuracy of localization for a long corridor according to the RGB-D camera and the 2D LiDAR. Full article

In recent years, simultaneous localization and mapping with the fusion of LiDAR and vision fusion has gained extensive attention in the field of autonomous navigation and environment sensing. However, its limitations in feature-scarce (low-texture, repetitive structure) environmental scenarios and dynamic environments have prompted researchers to investigate the use of combining LiDAR with other sensors, particularly the effective fusion with vision sensors. This technique has proven to be highly effective in handling a variety of situations by fusing deep learning with adaptive algorithms. LiDAR excels in complex environments, with its ability to acquire high-precision 3D spatial information, especially when dealing with complex and dynamic environments with high reliability. This paper analyzes the research status, including the main research results and findings, of the early single-sensor SLAM technology and the current stage of LiDAR and vision fusion SLAM. Specific solutions for current problems (complexity of data fusion, computational burden and real-time performance, multi-scenario data processing, etc.) are examined by categorizing and summarizing the body of the extant literature and, at the same time, discussing the trends and limitations of the current research by categorizing and summarizing the existing literature, as well as looks forward to the future research directions, including multi-sensor fusion, optimization of algorithms, improvement of real-time performance, and expansion of application scenarios. This review aims to provide guidelines and insights for the development of SLAM technology for LiDAR and vision fusion, with a view to providing a reference for further SLAM technology research. Full article

This study examines the application of low-cost 1D LiDAR sensors in drone-based stockpile volume estimation, with a focus on indoor environments. Three approaches were experimentally investigated: (i) a multi-drone system equipped with static, downward-facing 1D LiDAR sensors combined with an adaptive formation control algorithm; (ii) a single drone with a static, downward-facing 1D LiDAR following a zigzag trajectory; and (iii) a single drone with an actuated 1D LiDAR in an oscillatory fashion to enhance scanning coverage while following a shorter trajectory. The adaptive formation control algorithm, newly developed in this study, synchronises the drones’ waypoint arrivals and facilitates smooth transitions between dynamic formation shapes. Real-world experiments conducted in a motion-tracking indoor facility confirmed the effectiveness of all three approaches in accurately completing scanning tasks, as per intended waypoints allocation. A trapezoidal prism stockpile was scanned, and the volume estimation accuracy of each approach was compared. The multi-drone system achieved an average volumetric error of 1.3%, similar to the single drone with a static sensor, but with less than half the flight time. Meanwhile, the actuated LiDAR system required shorter paths but experienced a higher volumetric error of 4.4%, primarily due to surface reconstruction outliers and common LiDAR bias when scanning at non-vertical angles. Full article

Field observations provide valuable information for rockfall assessments, but estimating physical and statistical quantities related to rockfall propagation directly is challenging. Simulations are commonly used to infer these quantities, but their subjectivity can result in varying hazard land use zonation extents for different projects. This paper focuses on the application of simulated trajectories for rockfall hazard assessments, with an emphasis on reducing subjectivity. A quantitative guiding rockfall hazard methodology based on earlier concepts is presented and put in the context of legislated requirements. It details how the temporal hazard component, related to the likelihood of failure, can be distributed spatially using simulated trajectories. The method can be applied with results from any process-based software and combined with various prediction methods of the temporal aspect, although this aspect is not the primary focus. Applied examples for static objects and moving objects, such as houses and vehicles, are shown to illustrate the important effect of the object size. For that purpose, the methodology was applied at an indicative level over Norway utilizing its 1 m detailed digital terrain model (DTM) acquired from airborne LiDAR. Potential rockfall sources were distributed in 3D where slopes are steeper than 50°, as most rockfall events in the national landslide database (NSDB) occurred in such areas. This threshold considerably shifts toward gentler slopes when repeating the analysis with coarser DTMs. Simulated trajectories were produced with an adapted version of the simulation model stnParabel. Comparing the number of trajectories reaching the road network to the numerous related registered rockfall events of the NSDB, an indicative averaged yearly frequency of released rock fragments of 1/25 per 10,000 m² of cliff was obtained for Norway. This average frequency can serve as a starting point for hazard assessments and should be adjusted to better match local conditions. Full article

In recent years, pseudo point clouds generated from depth completion of RGB images and LiDAR data have provided a robust foundation for multimodal 3D object detection. However, the generation process often introduces noise, reducing data quality and detection accuracy. Moreover, existing methods fail to effectively capture channel correlations and global contextual information during the 2D feature extraction stage after the 3D backbone network, limiting detection performance. To address these challenges, this paper proposes NRAP-RCNN, a pseudo point cloud-based 3D object detection method with two key innovations: (1) A noise-reduction sparse convolution network (NRConvNet), comprising NRConv (noise-resistant submanifold sparse convolution), SRB (sparse convolution residual block), and MHSA (multi-head self-attention). NRConv suppresses pseudo point cloud noise by jointly encoding 2D and 3D features, SRB enhances feature extraction depth and robustness, and MHSA optimizes global feature representation. (2) An attention fusion module (ECA_GCA) is introduced to enhance the feature representation of the 2D backbone network by combining channel and global contextual information. The experimental results demonstrate that NRAP-RCNN achieves 88.4% car AP (R40) on the KITTI validation set and 85.1% on the test set, significantly outperforming advanced 3D detection methods, showcasing its effectiveness in improving detection performance. Full article

Pluvial flooding, driven by increasingly impervious surfaces and intense storm events, presents a growing challenge for urban areas worldwide. In Baltimore City, MD, USA, climate change, rapid urbanization, and aging stormwater infrastructure are exacerbating flooding impacts, resulting in significant socio-economic consequences. This study evaluated the effectiveness of a soil profile rehabilitation scenario using a 2D hydrodynamic modeling approach for the Tiffany Run watershed, Baltimore City. This study utilized different extreme storm events, a high-resolution (1 m) LiDAR Digital Terrain Model (DTM), building footprints, and hydrological soil data. These datasets were integrated into a fully coupled 2D hydrodynamic model, the City Catchment Analysis Tool (CityCAT), to simulate urban flood dynamics. The pre-soil rehabilitation simulation revealed a maximum water depth of 3.00 m in most areas, with hydrologic soil groups C and D, especially downstream of the study area. The post-soil rehabilitation simulation was targeted at vacant lots and public parcels, accounting for 33.20% of the total area of the watershed. This resulted in a reduced water depth of 2.50 m. Additionally, the baseline runoff coefficient of 0.49 decreased to 0.47 following the rehabilitation, and the model consistently recorded a peak runoff reduction rate of 4.10 across varying rainfall intensities. The validation using a contingency matrix demonstrated true-positive rates of 0.75, 0.50, 0.64, and 0 for the selected events, confirming the model’s capability at capturing real-world flood occurrences. Full article

Recent advances in autonomous driving have led to an increased use of LiDAR (Light Detection and Ranging) sensors for high-frequency 3D perceptions, resulting in massive data volumes that challenge in-vehicle networks, storage systems, and cloud-edge communications. To address this issue, we propose a bounded-error LiDAR compression framework that enforces a user-defined maximum coordinate deviation (e.g., 2 cm) in the real-world space. Our method combines multiple compression strategies in both axis-wise metric Axis or Euclidean metric L2 (namely, Error-Bounded Huffman Coding (EB-HC), Error-Bounded 3D Compression (EB-3D), and the extended Error-Bounded Huffman Coding with 3D Integration (EB-HC-3D)) with a lossless Huffman coding baseline. By quantizing and grouping point coordinates based on a strict threshold (either axis-wise or Euclidean), our method significantly reduces data size while preserving the geometric fidelity. Experiments on the KITTI dataset demonstrate that, under a 2 cm bounded-error, our single-bin compression reduces the data to 25–35% of their original size, while multi-bin processing can further compress the data to 15–25% of their original volume. An analysis of compression ratios, error metrics, and encoding/decoding speeds shows that our method achieves a substantial data reduction while keeping reconstruction errors within the specified limit. Moreover, runtime profiling indicates that our method is well-suited for deployment on in-vehicle edge devices, thereby enabling scalable cloud-edge cooperation. Full article

Structural changes in savanna trees vary spatially and temporally because of both biotic and abiotic drivers, as well as the complex interactions between them. Given this complexity, it is essential to monitor and quantify woody structural changes in savannas efficiently. We implemented a non-destructive approach based on Terrestrial Laser Scanning (TLS) and Quantitative Structure Models (QSMs) that offers the unique advantage of investigating changes in complex tree parameters, such as volume and branch length parameters that have not been previously reported for savanna trees. Leaf-off multi-scan TLS point clouds were acquired during the dry season, using a Riegl VZ1000 TLS, in September 2015 and October 2019 at the Skukuza flux tower in Kruger National Park, South Africa. These three-dimensional (3D) data covered an area of 15.2 ha with an average point density of 4270 points/m² (0.015°) and 1600 points/m² (0.025°) for the 2015 and 2019 clouds, respectively. Individual tree segmentation was applied on the two clouds using the comparative shortest-path algorithm in LiDAR 360(v5.4) software. We reconstructed optimized QSMs and assessed tree structural parameters such as Diameter at Breast Height (DBH), tree height, crown area, volume, and branch length at individual tree level. The DBH, tree height, crown area, and trunk volume showed significant positive correlations (R² > 0.80) between scanning periods regardless of the difference in the number of points of the matched trees. The opposite was observed for total and branch volume, total number of branches, and 1st-order branch length. As the difference in the point densities increased, the difference in the computed parameters also increased (R² < 0.63) for a high relative difference. A total of 45% of the trees present in 2015 were identified in 2019 as damaged/felled (75 trees), and the volume lost was estimated to be 83.4 m³. The results of our study showed that volume reconstruction algorithms such as TreeQSMs and high-resolution TLS datasets can be used successfully to quantify changes in the structure of savanna trees. The results of this study are key in understanding savanna ecology given its complex and dynamic nature and accurately quantifying the gains and losses that could arise from fire, drought, herbivory, and other abiotic and biotic disturbances. Full article

The aboveground biomass (AGB) of individual trees is a critical indicator for assessing urban forest productivity and carbon storage. In the context of global warming, it plays a pivotal role in understanding urban forest carbon sequestration and regulating the global carbon cycle. Recent advances in light detection and ranging (LiDAR) have enabled the detailed characterization of three-dimensional (3D) structures, significantly enhancing the accuracy of individual tree AGB estimation. This review examines studies that use LiDAR-derived 3D structural metrics to model and estimate individual tree AGB, identifying key metrics that influence estimation accuracy. A bibliometric analysis of 795 relevant articles from the Web of Science Core Collection was conducted using R Studio (version 4.4.1) and VOSviewer 1.6.20 software, followed by an in-depth review of 80 papers focused on urban forests, published after 2010 and selected from the first and second quartiles of the Chinese Academy of Sciences journal ranking. The results show the following: (1) Dalponte2016 and watershed are more widely used among 2D raster-based algorithms, and 3D point cloud-based segmentation algorithms offer greater potential for innovation; (2) tree height and crown volume are important 3D structural metrics for individual tree AGB estimation, and biomass indices that integrate these parameters can further improve accuracy and applicability; (3) machine learning algorithms such as Random Forest and deep learning consistently outperform parametric methods, delivering stable AGB estimates; (4) LiDAR data sources, point cloud density, and forest types are important factors that significantly affect the accuracy of individual tree AGB estimation. Future research should emphasize deep learning applications for improving point cloud segmentation and 3D structure extraction accuracy in complex forest environments. Additionally, optimizing multi-sensor data fusion strategies to address data matching and resolution differences will be crucial for developing more accurate and widely applicable AGB estimation models. Full article