High-definition (HD) mapping is essential for autonomous driving and localization services, providing detailed lane-level road graphs for various applications. Current methodologies primarily segment the geometric structure of lane lines from remote sensing images and extract vectorized road graphs using heuristic methods. However, these approaches fail to adequately account for lane instance information and topological structures. Furthermore, the semi-automated process imposes constraints on the spatial scalability of HD maps. To overcome these limitations, we propose AutoLRG, a two-stage method for lane-level road graph construction. In lane geometry prediction, we propose a lane segmentation network based on directional supervision and multimodal fusion, incorporating an angle-direction loss and a cross-attention-based fusion module to enhance lane perception and connectivity. In lane instance modeling, we develop a Transformer-based lane decoder, which leverages an object detection architecture to extract vectorized lane instances and road vertices in an end-to-end manner. In lane topology construction, we introduce a "road segment–intersection" decoupled model, which establishes the connectivity relationships of intersection nodes based on traffic regulations to form a lane-level topological directed road graph. The ablation studies conducted on the two benchmark datasets (UrbanLaneGraph and OpenSatMap) have validated the effectiveness of the method. Comparative experiments with other methods demonstrate that our approach exhibits superior performance in lane segmentation, instance modeling, and topology construction. Code is available at https://github.com/EchoQiHeng/AutoLRG.

Parallel computing on distributed heterogeneous architectures (e.g. computing clusters with multiple CPUs and GPUs) can significantly improve the computational efficiency and scalability of complicated algorithms, but it is theoretically and technically complex. Parallel raster processing libraries reduce the development complexity of parallel raster algorithms by hiding parallel computing details; however, no existing library sufficiently utilizes distributed heterogeneous computing resources. A general-purpose raster processing library (mcRPL) combining multi-process parallelism and multi-thread parallelism is proposed to enable parallel raster processing on distributed heterogeneous architectures with multiple CPUs and GPUs. Additionally, an adaptive hardware assignment strategy is proposed to fully utilize available processors in various hardware environments. A series of task-processing strategies are adopted to aim toward maximizing the utilization of the computing capacity of involved processors. Experiments revealed that two raster algorithms parallelized using mcRPL for spatiotemporal data fusion and land-use change simulation were 170.7- and 143.2-fold faster than original serial algorithms using 8 and 16 GPUs, respectively. While hiding the details of mixed parallelism and reducing the development complexity, mcRPL provides user-friendly interfaces for the development of parallel raster algorithms to enhance computational performance and enable large-scale raster computing tasks with extensive data volumes.

Existing traffic monitoring approaches cannot completely cover all road segments in real-time, leading to massive amounts of missing traffic data, which limits the implementation of intelligent transportation systems. Most existing methods lack deep mining of the unique spatiotemporal characteristics of traffic flows, resulting in difficulty in application to urban traffic with complex topologies and variable states. In this paper, we propose a novel Spatio-Temporal constrained Low-Rank Tensor Completion (ST-LRTC) method, which adopts a manifold embedding approach to depict the local geometric structure of spatiotemporal domains. Specifically, under the low-rank assumption, the method introduces temporal constraints based on the continuity and periodicity of traffic flow and a spatial constraint matrix reflecting the traffic flow transmission mechanism. We embed low-dimensional spatiotemporal constraint matrices into the low-rank tensor completion solving process to fully utilize the global features and local spatiotemporal characteristics of the traffic tensor. Experiments were performed using traffic data from Xi’an, China, and the results indicated that ST-LRTC outperformed state-of-the-art methods under various missing rates and patterns. Thorough experiments have demonstrated that the incorporation of spatiotemporal analysis can enhance the adaptability of the tensor completion model to complex urban scenarios, which guarantees better monitoring, diagnosis, and optimization of urban traffic states.

Xue Yanga, Kathleen Stewartb, Mengyuan Fangc & Luliang Tangc*a School of Geography and Information Engineering, China University of Geosciences, Wuhan, Chinab Department of Geographical Sciences, University of Maryland, College Park, MD, USAc State Key Laboratory for Information Engineering in Surveying, Mapping and Remote Sensing, Wuhan University, Wuhan, ChinaXue Yang received the Ph.D. degree from Wuhan University, Wuhan, China, in 2018. She is currently an associate Professor with China University of Geosciences, Wuhan. Her research interests include intelligent transportation system, spatiotemporal data analysis, and information mining. Homepage: http://grzy.cug.edu.cn/yangxue1/zh_CN/index.htmEmail: Kathleen Stewart is currently a Professor in the Department of Geographical Sciences and Director of the Center for Geospatial Information Science. She works in the area of geographic information science with a particular focus on geospatial dynamics. She is interested in mobility and spatial access, often in a big geospatial data context and using approaches that lie in the expanding field of spatial data science. Homepage: https://geog.umd.edu/facultyprofile/stewart/kathleenEmail: Mengyuan Fang received Bsc degree from Wuhan University, Wuhan, China, 2014. He is currently a Ph.D candidate at the State Key Laboratory of Information Engineering in Surveying, Mapping and Remote Sensing, Wuhan University. His research addresses the issue of traffic congestion detection and prediction using big trace data.Email: Luliang Tang received the Ph.D. degree from Wuhan University, Wuhan, China, in 2007. He is currently a Professor with Wuhan University. His research interests include space–time GIS, GIS for transportation, and change detection. Homepage: http://www.lmars.whu.edu.cn/index.php/js/298.htmlEmail: CONTACT Luliang Tang [email protected] lack of associating pedestrian networks, i.e. the paths and roads used for non-vehicular travel, with information about semantic attribution is a major weakness for many applications, especially those supporting accurate pedestrian routing. Researchers have developed various algorithms to generate pedestrian walkways based on datasets, including high-resolution images, existing map databases, and GPS data; however, the semantic attribution of pedestrian walkways is often ignored. The objective of our study is to automatically extract semantic information including incline values and the different categories of pedestrian paths from multi-source spatial data, such as crowdsourced GPS tracking data, land use data, and motor vehicle road (MVR) networks. Incline values for each pedestrian path were derived from tracking data through elevation filtering using wavelet theory and a similarity-based map-matching method. To automatically categorize pedestrian paths into five classes including sidewalk, crosswalk, entrance walkway, indoor path, and greenway, we developed a hierarchical strategy of spatial analysis using land use data and MVR networks. The effectiveness of our proposed method is demonstrated using real datasets including GPS tracking data collected by volunteers, land use data acquired from OpenStreetMap, and MVR network data downloaded from Gaode Map.

Spatiotemporal data fusion is a cost-effective way to produce remote sensing images with high spatial and temporal resolutions using multisource images. Using spectral unmixing analysis and spatial interpolation, the flexible spatiotemporal data fusion (FSDAF) algorithm is suitable for heterogeneous landscapes and capable of capturing abrupt land-cover changes. However, the extensive computational complexity of FSDAF prevents its use in large-scale applications and mass production. Besides, the domain decomposition strategy of FSDAF causes accuracy loss at the edges of subdomains due to the insufficient consideration of edge effects. In this study, an enhanced FSDAF (cuFSDAF) is proposed to address these problems, and includes three main improvements. First, the TPS interpolator is replaced by an accelerated inverse distance weighted (IDW) interpolator to reduce computational complexity. Second, the algorithm is parallelized based on the compute unified device architecture (CUDA), a widely used parallel computing framework for graphics processing units (GPUs). Third, an adaptive domain decomposition (ADD) method is proposed to improve the fusion accuracy at the edges of subdomains and to enable GPUs with varying computing capacities to deal with datasets of any size. Experiments showed while obtaining similar accuracies to FSDAF and an up-to-date deep-learning-based method, cuFSDAF reduced the computing time significantly and achieved speed-ups of 140.3–182.2 over the original FSDAF program. cuFSDAF is capable of efficiently producing fused images with both high spatial and temporal resolutions to support applications for large-scale and long-term land surface dynamics. Source code and test data available athttps://github.com/HPSCIL/cuFSDAF.

The Spatial–Temporal Savitzky–Golay (STSG) method for noise reduction can address the problem of temporally continuous Normalized Difference Vegetation Index (NDVI) gaps and effectively increase local low NDVI values without overcorrection. However, STSG largely depends on the quality flags of the NDVI time-series data, and inaccurate quality flags yield misleading final results. STSG also requires extensive computing time when used in large-scale applications. This study proposes an enhanced method, called cuSTSG, to address the aforementioned limitations of STSG. First, cosine similarities between the annual NDVI time series were used to identify and exclude the NDVI values with inaccurate quality flags from the NDVI seasonal growth trajectory. Second, computational performance was improved by reducing redundant computations and parallelizing computationally intensive procedures using the Compute Unified Device Architecture (CUDA) on graphics processing units (GPUs). Experiments on four MODIS NDVI time-series datasets of various sizes and regions showed that compared with the original STSG, cuSTSG reduced the mean absolute errors of the final products by 4.90%, 7.77%, 11.76% and 2.06%, respectively. The results also showed that cuSTSG on a GPU achieved 75+ speed-up compared with the Interactive Data Language-implemented STSG, and 30+ speed-up compared with the C++-implemented STSG. cuSTSG can effectively mitigate the impacts of inaccurate quality flags on final products and generate high-quality NDVI time series at large scales with high accuracy and performance. The source code of cuSTSG is available at https://github.com/HPSCIL/cuSTSG.

Short-term traffic prediction is of great importance to the management of traffic congestion, a pervasive and difficult-to-solve problem in many metropolises all over the world. However, existing studies on traffic prediction contain rough traffic information at the carriageway level that ignore the distinction between different turns in one intersection. With the aim of predicting traffic at road intersections from big trace data on a finer scale, this study proposes a novel method, the fine-grained traffic prediction method (FTPG) with a graph attention network (GAT), which predicts traffic information, including traffic flow speeds, traffic states, and average queue lengths, at the turn level. In the FTPG, a method for estimation of the queue starting point is proposed to improve the accuracy of traffic information detection. Furthermore, the topology is constructed under turn-level conditions, and a GAT-based method, the spatio-temporal residual graph attention network (ST-RGAN), is proposed to improve the prediction accuracy. Experiments are performed using taxi GPS trace data collected in the city of Wuhan and show that the proposed FTPG method can make predictions with fine-grained traffic information for road intersections accurately and robustly.

As an important component in transportation maps, three-dimensional (3D) structure information of grade-separated junctions is crucial for applications such as intelligent driving, route planning and traffic control. In order to acquire spatial layouts of road junctions, researchers have developed algorithms to extract planar structures from various data sources. However, it is less common to refine maps of grade-separated junctions with 3D structure information using tracking data. The objective of this study is to find an approach to extracting 3D structures of grade-separated junctions from vehicle trajectories. The proposed method is based on semantic segmentation and data fusion. Trajectories were divided into sections with different trends of elevation by detecting change points. The ranges and elevations of slopes and level sections were derived by seeking consensus among different trajectories using a data fusion technique. Based on semantic segmentation and aggregated elevations, we reconstructed detailed 3D junction structures. This method was validated on multiple crowdsourced trajectory datasets and compared to cluster center linking method. Experiments show that the proposed method had a higher overall accuracy of semantic segmentation than baseline method. The accuracy of vertical relationship at intersections is comparable to baseline. Despite large elevation discrepancy among trajectories, the performance of the proposed method was similar across crowdsourced trajectory datasets from open and commercial projects.

Pedestrian networks play an important role in various applications, such as pedestrian navigation services and mobility modeling. This paper presents a novel method to extract pedestrian networks from crowdsourced tracking data based on a two-layer framework. This framework includes a walking pattern classification layer and a pedestrian network generation layer. In the first layer, we propose a multi-scale fractal dimension (MFD) algorithm in order to recognize the two different types of walking patterns: walking with a clear destination (WCD) or walking without a clear destination (WOCD). In the second layer, we generate the pedestrian network by combining the pedestrian regions and pedestrian paths. The pedestrian regions are extracted based on a modified connected component analysis (CCA) algorithm from the WOCD traces. We generate the pedestrian paths using a kernel density estimation (KDE)-based point clustering algorithm from the WCD traces. The pedestrian network generation results using two actual crowdsourced datasets show that the proposed method has good performance in both geometrical correctness and topological correctness.

Lane-level road network updating is crucial for urban traffic applications that use geographic information systems contributing to, for example, intelligent driving, route planning and traffic control. Researchers have developed various algorithms to update road networks using sensor data, such as high-definition images or GPS data; however, approaches that involve change detection for road networks at lane level using GPS data are less common. This paper presents a novel method for automatic change detection of lane-level road networks based on GPS trajectories of vehicles. The proposed method includes two steps: map matching at lane level and lane-level change recognition. To integrate the most up-to-date GPS data with a lane-level road network, this research uses a fuzzy logic road network matching method. The proposed map-matching method starts with a confirmation of candidate lane-level road segments that use error ellipses derived from the GPS data, and then computes the membership degree between GPS data and candidate lane-level segments. The GPS trajectory data is classified into successful or unsuccessful matches using a set of defuzzification rules. Any topological and geometrical changes to road networks are detected by analysing the two kinds of matching results and comparing their relationships with the original road network. Change detection results for road networks in Wuhan, China using collected GPS trajectories show that these methods can be successfully applied to detect lane-level road changes including added lanes, closed lanes and lane-changing and turning rules, while achieving a robust detection precision of above 80%.

Kernel Density Estimation (KDE) is an important approach to analyse spatial distribution of point features and linear features over 2-D planar space. Some network-based KDE methods have been developed in recent years, which focus on estimating density distribution of point events over 1-D network space. However, the existing KDE methods are not appropriate for analysing the distribution characteristics of certain kind of features or events, such as traffic jams, queue at intersections and taxi carrying passenger events. These events occur and distribute in 1-D road network space, and present a continuous linear distribution along network. This paper presents a novel Network Kernel Density Estimation method for Linear features (NKDE-L) to analyse the space–time distribution characteristics of linear features over 1-D network space. We first analyse the density distribution of each linear feature along networks, then estimate the density distribution for the whole network space in terms of the network distance and network topology. In the case study, we apply the NKDE-L to analyse the space–time dynamics of taxis’ pick-up events, with real road network and taxi trace data in Wuhan. Taxis’ pick-up events are defined and extracted as linear events (LE) in this paper. We first conduct a space–time statistics of pick-up LE in different temporal granularities. Then we analyse the space–time density distribution of the pick-up events in the road network using the NKDE-L, and uncover some dynamic patterns of people’s activities and traffic condition. In addition, we compare the NKDE-L with quadrat method and planar KDE. The comparison results prove the advantages of the NKDE-L in analysing spatial distribution patterns of linear features in network space.

Lane-based road network information, such as the number and locations of traffic lanes on a road, has played an important role in intelligent transportation systems. In this paper, we propose a Collecting Lane-based Road Information via Crowdsourcing (CLRIC) method, which can automatically extract detailed lane structure of roads by using crowdsourcing data collected by vehicles. First, CLRIC filters the high-precision GPS data from the raw trajectories based on region growing clustering with prior knowledge. Second, CLRIC mines the number and locations of traffic lanes through optimized constrained Gaussian mixture model. Experiments are conducted with taxi GPS trajectories in Wuhan, China, and the results show that CLRIC is quantified and displays detailed road networks with the number and locations of traffic lanes comparing with the satellite image and human-interpreted situation.