Map Generation Method Research Articles

Exploring BTI aging effects on spatial power density and temperature profiles of VLSI chips

The Long-term reliability of a chip, encompassing factors like bias temperature instability (BTI), plays a substantial role in the chip's operational efficiency and overall lifespan. Most studies primarily center around performance-related aspects like delay and timing impacts, and fewer studies are performed on reliability impacts on the spatial power density and thermal profiles of the chips. In this study, we propose to investigate the BTI impacts on the spatial power density and temperature profiles of VLSI chips for the first time. We assessed the BTI aging impact on the on-chip spatial power density and temperature for two widely used circuit functional blocks (dual port RAM, Discrete Cosine Transform (DCT) block) at T = 130◦C and T = 25◦C to account for the worst-case BTI degradation, using degradation-aware cell libraries for a 10-year aging scenario. Furthermore, we showcased the essential role of BTI aging-aware timing analysis in evaluating the impact of BTI aging on total power, on-chip spatial power density, and thermal maps. Neglecting this aspect can result in a substantial underestimation of the results related to the parameters mentioned above. We developed a power map generation method from the circuit layout and power analysis from EDA tools. We demonstrate that both circuits’ maximum power density reduction is approximately 12 % and 20 %, respectively. Furthermore, to analyze the BTI impact on spatial temperature, we built the heat transfer model using a multiphysics tool to imitate a real chip (Intel i7-8650U) and performed thermal simulations to evaluate the spatial thermal map. The resulting maximum temperature reduction for both these circuits is approximately 10 % and 12 %, respectively, which is quite significant.Our analysis has further unveiled that, in the context of a specific circuit, the position of maximum power density and the occurrence of a hot spot remains consistent over time, unaffected by aging. However, it's important to note that these positions can vary between different circuits, primarily influenced by the workload the circuit is currently handling. Furthermore, our findings demonstrate that the effects of Bias Temperature Instability (BTI) aging are significantly more pronounced when the circuit operates at higher temperatures (T = 130◦C) compared to lower operating temperatures (T = 25◦C).

Model-agnostic progressive saliency map generation for object detector

With the widespread adoption of object detection models across various industries, the interpretability of these detectors has become an important research topic. The interpretability of a detector helps humans understand which areas significantly contribute to the model's decision. Furthermore, the interpretability enhances the credibility of detectors and helps identify their strengths and weaknesses. Due to the ability to provide intuitive explanations of models, the saliency map has been widely employed in the field of interpreting deep models. Model-agnostic interpretability methods are more general approaches as they treat the model as a black box without considering its internal complexity structure. However, existing model-agnostic interpretability methods often introduce “noise” into saliency maps by applying random masking and fixed masking granularity. This noise reduces the quality and interpretability of the generated saliency maps. To address this challenge and obtain more interpretable saliency maps for object detection models, this paper proposes a model-agnostic progressive saliency map generation method based on a hierarchical framework called MAPSM. In MAPSM, an adaptive masking partition mechanism is introduced to adapt the masking granularity to different object sizes. Additionally, MAPSM employs a saliency-driven mask generation strategy to effectively reduce the “noise”. Utilizing a hierarchical framework, MAPSM progressively discovers and refines the saliency areas of objects, resulting in more interpretable saliency maps. To evaluate the quality of the saliency maps generated by MAPSM, we compare it with other methods in multiple metrics. Experimental results demonstrate that our method produces saliency maps with better quality and interpretability.

[formula omitted]D-CS: A circularly shifting chaotic map generation method

Due to more complex behavior and larger number of control parameters, high-dimensional chaotic map generation methods may provide more satisfactory performance in various practical applications compared to low-dimensional counterparts. Designing n-dimensional (nD) chaotic map generation methods for generating chaotic maps in arbitrary dimensions with desired dynamics is an interesting but challenging task. In this paper, we propose a universal framework called the nD circularly shifting chaotic map generation method (nD-CS), which utilizes existing one-dimensional chaotic maps as seed maps to generate nD chaotic maps with complex and robust behaviors. Theoretical analysis proves that when specific criteria are met by its control parameters, our nD-CS can exhibit complex and robust hyperchaotic behavior in the sense of Lyapunov exponent. To demonstrate the effectiveness of our nD-CS, we first employ it to generate three examples of three-dimensional hyperchaotic maps. The results indicate high-performance indicators of these new chaotic maps. Furthermore, we compare the nD chaotic maps generated by our nD-CS to those produced by existing methods. The results demonstrate that our chaotic maps have a better overall performance.

Satellite true digital orthophoto map generation without elevation data: a New NeRF-based method

ABSTRACT Existing satellite true digital orthophoto map (TDOM) generation methods are designed for ground elevation data. Obtaining large-scale and high-precision ground 3D data is prohibitively costly, while low-precision elevation data introduces issues such as relief displacement, boundary distortion, and artefacts. Furthermore, producing TDOMs from satellite images captured under various lighting conditions can cause colour inconsistency problems. These issues impose limitations on the application and development of satellite TDOMs. In this paper, we propose a novel image-to-image approach that directly generates high-quality TDOMs from multi-view satellite images without the need for elevation data as input. Specifically, the 3D scene is efficiently represented by the volume density and colour, which are modelled utilizing a neural network. During each iteration, this 3D representation undergoes optimization by the multi-view satellite signals, employing a volume rendering formula. Finally, TDOM is produced utilizing our true ortho-volume rendering technique. Experimental results demonstrate that our TDOM achieves superior visual quality and geometry accuracy without the need for 3D elevation data.

Waypoint Transfer Module between Autonomous Driving Maps Based on LiDAR Directional Sub-Images

Lane graphs are very important for describing road semantics and enabling safe autonomous maneuvers using the localization and path-planning modules. These graphs are considered long-life details because of the rare changes occurring in road structures. On the other hand, the global position of the corresponding topological maps might be changed due to the necessity of updating or extending the maps using different positioning systems such as GNSS/INS-RTK (GIR), Dead-Reckoning (DR), or SLAM technologies. Therefore, the lane graphs should be transferred between maps accurately to describe the same semantics of lanes and landmarks. This paper proposes a unique transfer framework in the image domain based on the LiDAR intensity road surfaces, considering the challenging requirements of its implementation in critical road structures. The road surfaces in a target map are decomposed into directional sub-images with X, Y, and Yaw IDs in the global coordinate system. The XY IDs are used to detect the common areas with a reference map, whereas the Yaw IDs are utilized to reconstruct the vehicle trajectory in the reference map and determine the associated lane graphs. The directional sub-images are then matched to the reference sub-images, and the graphs are safely transferred accordingly. The experimental results have verified the robustness and reliability of the proposed framework to transfer lane graphs safely and accurately between maps, regardless of the complexity of road structures, driving scenarios, map generation methods, and map global accuracies.

STRATEGY ON HIGH-DEFINITION POINT CLOUD MAP CREATION FOR AUTONOMOUS DRIVING IN HIGHWAY ENVIRONMENTS

Abstract. In recent years, a lot of researchers have been trying the development of efficient ways to create HD maps with centimeter-level precision. Mobile mapping system (MMS) produce 3D HD point cloud map of the surrounding by integrating navigation (i.e., direct georeferencing or DG process) and high-resolution imaging sensor data. Unfortunately, in partially environments, the provided accuracy of the GNSS system degrades dramatically. In order to constraint the drift and correct the georeferenced point cloud map, ground control points (GCPs) are placed along the road. Moreover, there are approaches which use laser-based point cloud registration techniques to construct the point cloud map. However, all promising mapping techniques which currently use the laser as the core sensor for mapping the high-definition point cloud map may not be promised to construct the point cloud map in partially or unfriendly environments. As a literature review and result, a suitable approach for creating the promising point cloud map is to combine the INS/GNSS navigation solution, LiDAR matching techniques, and GCPs. Thus, this study introduces the HD point cloud map generation method that can potentially help researchers create personalized and globalized HD point cloud maps and develop new HD point cloud map generation methodologies.

Augmented Grad-CAM++: Super-Resolution Saliency Maps for Visual Interpretation of Deep Neural Network

In recent years, deep neural networks have shown superior performance in various fields, but interpretability has always been the Achilles’ heel of deep neural networks. The existing visual interpretation methods for deep neural networks still suffer from inaccurate and insufficient target localization and low-resolution saliency maps. To address the above issues, this paper presents a saliency map generation method based on image geometry augmentation and super-resolution called augmented high-order gradient weighting class activation mapping (augmented grad-CAM++). Unlike previous approaches that rely on a single input image to generate saliency maps, this method first introduces the image geometry augmentation technique to create a set of augmented images for the input image and generate activation mappings separately. Secondly, the augmented activation mappings are combined to form the final saliency map. Finally, a super-resolution technique is introduced to add pixel points to reconstruct the saliency map pixels to improve the resolution of the saliency map. The proposed method is applied to analyze standard image data and industrial surface defect images. The results indicate that, in experiments conducted on standard image data, the proposed method achieved a 3.1% improvement in the accuracy of capturing target objects compared to traditional methods. Furthermore, the resolution of saliency maps was three times higher than that of traditional methods. In the application of industrial surface defect detection, the proposed method demonstrated an 11.6% enhancement in the accuracy of capturing target objects, concurrently reducing the false positive rate. The presented approach enables more accurate and comprehensive capture of target objects with higher resolution, thereby enhancing the visual interpretability of deep neural networks. This improvement contributes to the greater interpretability of deep learning models in industrial applications, offering substantial performance gains for the practical deployment of deep learning networks in the industrial domain.

Ground truth based comparison of saliency maps algorithms

Deep neural networks (DNNs) have achieved outstanding results in domains such as image processing, computer vision, natural language processing and bioinformatics. In recent years, many methods have been proposed that can provide a visual explanation of decision made by such classifiers. Saliency maps are probably the most popular. However, it is still unclear how to properly interpret saliency maps for a given image and which techniques perform most accurately. This paper presents a methodology to practically evaluate the real effectiveness of saliency map generation methods. We used three state-of-the-art network architectures along with specially prepared benchmark datasets, and we proposed a novel metric to provide a quantitative comparison of the methods. The comparison identified the most reliable techniques and the solutions which usually failed in our tests.

Weakly Supervised Concept Map Generation through Task-Guided Graph Translation.

Recent years have witnessed the rapid development of concept map generation techniques due to their advantages in providing well-structured summarization of knowledge from free texts. Traditional unsupervised methods do not generate task-oriented concept maps, whereas deep generative models require large amounts of training data. In this work, we present GT-D2G (Graph Translation-based Document To Graph), an automatic concept map generation framework that leverages generalized NLP pipelines to derive semantic-rich initial graphs, and translates them into more concise structures under the weak supervision of downstream task labels. The concept maps generated by GT-D2G can provide interpretable summarization of structured knowledge for the input texts, which are demonstrated through human evaluation and case studies on three real-world corpora. Further experiments on the downstream task of document classification show that GT-D2G beats other concept map generation methods. Moreover, we specifically validate the labeling efficiency of GT-D2G in the label-efficient learning setting and the flexibility of generated graph sizes in controlled hyper-parameter studies.

Lane-Level Map Generation and Management Framework Using Connected Car Data

This study proposes a lane-level map generation and management framework using connected sensor data to reduce the manpower and time required for producing and updating high-definition (HD) maps. Unlike previous studies that relied on the onboard processing capabilities of vehicles to collect map-constructing elements, this study offloads computing for map generation to the cloud, assigning vehicles solely the role of transmitting sensor data. For efficient data collection, we divide the space into a grid format to define it as a partial map and establish the state of each map and its transition conditions. Lastly, tailored to the characteristics of the road elements composing the map, we propose an automated map generation technique and method for selectively collecting data. The map generation method was tested using data collected from actual vehicles. By transmitting images with an average size of 350 KB, implementation was feasible even with the current 5G upload bandwidth. By utilizing 12,545 elements, we were able to achieve a position accuracy and regression RMSE of less than 0.25 m, obtaining 651 map elements to construct the map. We anticipate that this study will help reduce the manpower and time needed for deploying and updating HD maps.

VR technology used to show the practice of architectural node structure

With the development and maturity of virtual reality (VR for short) and artificial intelligence technology, panoramic VR roaming is applied to more and more industries. In the construction engineering industry, due to the complex structure of some key building nodes, the application of panoramic VR roaming technology modeling will play a role in significantly reducing the workload and understanding difficulty of design, construction and management personnel. Therefore, this study improves the parallax map generation method of the belief propagation (BP) algorithm in the panoramic VR roaming to use the form of limit matching and optimizes the energy function and matching primitives in it, so as to propose a spatial model construction method of the panoramic VR roaming based on the improved BP algorithm. The experimental results show that the panoramic VR roaming space model construction method proposed in this study can significantly improve the quality of spatial modeling of building node images, and the normalized values of structural similarity of VR roaming space models designed based on Improved belief propagation (IBP), GC, BP, and DP algorithms are 3.32, 3.23, 2.96, and 2.84, respectively, when the number of iterations is obtained 200 times, is also the highest among the compared methods, for example, when the number of samples is 400, the calculation time of IBP scheme is 14.73 ± 0.85 min, and the highest time of other three schemes is 6.86 ± 0.67 min. Therefore, this virtual space modeling method is designed to be more suitable for application scenarios that require lower computational timeliness.

A Socially-Aware Radio Map Framework for Improving QoS of UAV-Assisted MEC Networks

The expeditious growth of the Internet of Things (IoT) has accelerated the evolution of multi-access edge computing (MEC). MEC alleviates the challenges of conventional cloud computing, such as high data latency, poor data gathering reliability, increased network cost, and lack of network robustness. The primary objective of MEC is to facilitate a hierarchy of edge servers to address these quality-of-service (QoS) challenges, especially the information propagation issue due to the mobility of IoT devices (IoD). Further, social-relationship among mobile IoD is a critical parameter used to reduce the data transmission delay and queue size at the MEC. Specifically, in this work, a novel socially-aware radio map generation method is proposed to compute the fine-grained and accurate locations of QoS-deprived areas. Firstly, a novel method to compute the social relationship index (SRI) factor is proposed on the basis of current and future encounters among moving IoDs. Then the obtained SRI factor is used to form clusters of mobile IoD. The clusters’ signal to interference plus noise ratio (SINR) is then used to generate the socially-aware radio map. Following that, unmanned aerial vehicles (UAV) use this radio map, which contains rich and serviceable channel information, for 3D beamforming towards the mobile clusters. Using the obtained radio map, Kalman filter-based offline path planning of UAVs is proposed to minimize the UAVs flying distance from the initial to final locations. Furthermore, an optimization problem is formulated to assess the performance of the proposed method. Finally, the performance of the proposed method is compared with the existing methods, taking into account various network parameters such as optimum number of UAVs needed to cover the deployed area, data transmission delay, and received SINR.

Using neural networks to create a reliable phase quality map for phase unwrapping.

Two-dimensional phase unwrapping is a crucial step in interferometric signal processing. A phase quality map can help the unwrapping algorithm deal with low-quality and fast-changing regions. However, because existing algorithms cannot calculate a quality map representing the gradient quality directly, it is usually necessary to approximate the gradient quality with phase quality to assist the network-based phase unwrapping algorithm. Furthermore, they cannot withstand intense noise in low-quality regions, resulting in many errors in path-based algorithms. To address the aforementioned issues, this paper analyzes the essence of a quality map and proposes a quality map generation method based on a convolutional neural network. The generated quality maps are a pair, each indicating the quality of horizontal and vertical gradients. Experiments show that the quality map generated by this method can help path-based and network-based algorithms perform better.

A Novel Spatio–Temporal Deep Learning Vehicle Turns Detection Scheme Using GPS-Only Data

Whether the computer is driving your car or you are, advanced driver assistance systems (ADAS) come into play on all levels, from weather monitoring to safety. These modern-day ADASs use various assisting tools for drivers to keep the journey safe; these sophisticated tools provide early signals of numerous events, such as road conditions, emerging traffic scenarios, and weather warnings. Many urban applications, such as car-sharing and logistics, rely on accurate and up-to-date road map data. Map generation methods use a variety of data sources, including but not limited to global positioning systems (GPS). In this research we propose a GPS-only data trajectory analysis and a novel scheme to convert GPS trajectory data to image-based data to train a custom Convolutional Neural Network (CNN) model. The empirical results with an extensive 5-fold cross-validation show that the proposed scheme identifies turn and not turn with more than 94% recall. It outperforms the existing turn detection schemes on two major frontiers, the required data and the accuracy achieved in detecting different driving behaviors.

A Novel Data-Driven-Based Component Map Generation Method for Transient Aero-Engine Performance Adaptation

Accurate component maps, which can significantly affect the efficiency, reliability and availability of aero-engines, play a critical role in aero-engine performance simulation. Unfortunately, the information of component maps is insufficient, leading to substantial limitations in practical application, wherein compressors are of particular interest. Here, a data-driven-based compressor map generation approach for transient aero-engine performance adaptation is investigated. A multi-layer perceptron neural network is utilized in simulating the compressor map instead of conventional interpolation schemes, and an adaptive variable learning rate backpropagation (ADVLBP) algorithm is employed to accelerate the convergence and improve the stability in the training process. Aside from that, two different adaptation strategies designed for steady state and transient conditions are implemented to adaptively retrain the compressor network according to measurement deviations until the accuracy requirements are satisfied. The proposed method is integrated into a turbofan component-level model, and simulations reveal that the ADVLBP algorithm has the capability of more rapid convergence compared with conventional training algorithms. In addition, the maximum absolute measurement deviation decreased from 6.35% to 0.44% after steady state adaptation, and excellent agreement between the predictions and benchmark data was obtained after transient adaptation. The results demonstrate the effectiveness and superiority of the proposed component map generation method.

The BINGO project

Context.BINGO (Baryon Acoustic Oscillations from Integrated Neutral Gas Observations) is a radio telescope designed to survey from 980 MHz to 1260 MHz, observe the neutral hydrogen (H I) 21 cm line, and detect the baryon acoustic oscillation signal with the intensity mapping technique. Here we present our method for generating mock maps of the 21 cm intensity mapping signal that cover the BINGO frequency range and related test results.Aims.We would like to employN-body simulations to generate mock 21 cm intensity maps for BINGO and study the information contained in 21 cm intensity mapping observations about structure formation, H Idistribution and H Imass-halo mass relation.Methods.We fit an H Imass-halo mass relation from the ELUCID semianalytical galaxy catalog and applied it to the Horizen Run 4 halo catalog to generate the 21 cm mock map, which is called HOD. We also applied the abundance-matching method and matched the Horizen Run 4 galaxy catalog with the H Imass function measured from ALFALFA, to generate the 21 cm mock map, which is called HAM.Results.We studied the angular power spectrum of the mock maps and the corresponding pixel histogram. The comparison of two different mock map generation methods (HOD and HAM) is presented. We provide the fitting formula of ΩHi, H Ibias, and the lognormal fitting parameter of the maps, which can be used to generate similar maps. We discuss the possibility of measuring ΩHiand H Ibias by comparing the angular power spectrum of the mock maps and the theoretical calculation. We also discuss the redshift space distortion effect, the nonlinear effect, and the bin size effect in the mock map.Conclusions.By comparing the angular power spectrum measured from two different types of mock maps and the theoretical calculation, we find that the theoretical calculation can only fit the mock result at large scales. At small scales, neither the linear calculation nor the halofit nonlinear calculation can provide an accurate fitting, which reflects our poor understanding of the nonlinear distribution of H Iand its scale-dependent bias. We have found that the bias is highly sensitive to the method of populating H Iin halos, which also means that we can place constraints on the H Idistribution in halos by observing 21 cm intensity mapping. We also illustrate that only with thin frequency bins (such as 2 MHz), we can discriminate the Finger-of-God effect. All of our investigations using mocks provide useful guidance for our expectation of BINGO experiments and other 21 cm intensity mapping experiments.

Dual U-Net based feature map algorithm for automatic projection alignment of synchrotron nano-CT

Synchrotron radiation-based nano-CT is a non-destructive imaging technique used to study the internal structure of matter. The efficient and automatic correction of errors in nano-CT imaging has proven challenging. Accordingly, different projection alignment techniques have been employed to address these issues. However, these techniques have their limitations. Herein, we propose a novel error correction model, a U-Net, that combines an experimentally informed mathematical model, optimized deep learning algorithm, and efficient projection matching model. The results show that the generation of feature maps based on our model is effective for improving the quality of 3D reconstructions while remaining highly robust. We further propose an optimized feature map generation method, a dual U-Net. Applying it yields an almost negligible alignment error rate for low random error-shift nano-CT data and a stable decrease in errors of one order of magnitude for high random error-shift nano-CT data. This study demonstrates the feasibility of the proposed feature map replacement projection alignment method, which may prove crucial to maximizing the efficiency and generalization of automatic projection alignment tasks.

Improved YOLOv4 for Pedestrian Detection and Counting in UAV Images.

UAV (unmanned aerial vehicle) captured images have small pedestrian targets and loss of key information after multiple down sampling, which are difficult to overcome by existing methods. We propose an improved YOLOv4 model for pedestrian detection and counting in UAV images, named YOLO-CC. We used the lightweight YOLOv4 for pedestrian detection, which replaces the backbone with CSPDarknet-34, and two feature layers are fused by FPN (Feature Pyramid Networks). We expanded the perception field using multiscale convolution based on the high-level feature map and generated the population density map by feature dimension reduction. By embedding the density map generation method into the network for end-to-end training, our model can effectively improve the accuracy of detection and counting and make feature extraction more focused on small targets. Our experiments demonstrate that YOLO-CC achieves 21.76 points AP50 higher than that of the original YOLOv4 on the VisDrone2021-counting data set while running faster than the original YOLOv4.

Seabed Sediment Classification Using Spatial Statistical Characteristics

Conventional sediment classification methods based on Multibeam Echo System (MBES) data have low accuracy since the correlation between features and sediment has not been fully considered. Moreover, their poor resistance to the residual error of MBES backscatter strength (BS) processing also degrades their performances. Toward these problems, we propose a seabed sediment classification method using spatial statistical features extracted from angular response curve (ARC), topography, and geomorphology. First, to reduce interference of noise and residual error of beam pattern correction, we propose a robust method combining the Generic Seafloor Acoustic Backscatter (GSAB) model and Huber loss function to estimate the parameters of ARC which is strongly correlated with seabed sediments. Second, a feature set is constructed by AR features composed of GSAB parameters, BS mosaic and its derivatives, and seabed topography and its derivatives to characterize seabed sediments. After that, feature selection and probability map acquisition are employed based on the random forest algorithm (RF). Finally, a denoising and final sediment map generation method is proposed and applied to probability maps to obtain the sediment map with reasonable sediment distribution and clear boundaries between classes. We implement experiments and achieve the classification accuracy of 93.3%, which verifies the validity of our method.

Global path planning based on BIM and physics engine for UGVs in indoor environments

This paper proposes a global path planning (GPP) system based on building information modeling (BIM) and physics engine for unmanned ground vehicles (UGVs) operations in indoor environments. Firstly, UGV configuration is integrated into BIM as a knowledge base of GPP system by customized IFC structure, with which a multi-layer map generation method is proposed with improved logic and efficiency. Secondly, a UGV-centric A* path planning algorithm is designed by considering UGV's properties, including mobility-based primitive expansion, geometry-based collision checking, mobility-based cost setting, and mobility-based analytical expansion. Finally, a reliable trajectory generation method is developed based on physics engine, followed with a novel spatiotemporal coordination method for efficient collision avoidance. The whole GPP system is validated in a representative university building floor and a common inspection UGV. It is demonstrated that UGV-integrated BIM improves the resources interoperability for the whole system, and the map components are clarified clearly for efficient generation and maintenance. Besides, the UGV-centric A* performs a 100% success rate in congested environments where traditional A* always fail. It can even reduce almost 50% of the trajectory time and steering jerk in open scenarios than traditional A*, but still with an acceptable computation speed (less than 0.6 s). Finally, trajectory coordination saves 50% of path traveling time compared with the general queue method.