Road Centerline Research Articles

Road Centerline Data Accumulation for Rescue Workers Whose Expertise Is Not GIS.

The objective of this study is to provide road centerline data for professionals of disaster medicine areas who are often beginners in GIS use. Newly developed vector tile format data were converted into shapefile format data, then were organized as second level medical districts to which medical professionals are accustomed. Road centerline data in Japan is being prepared to release from Association for Promotion of Infrastructure Geospatial Information Distribution free of charge. Professionals of disaster medicine areas increased their accessibility of GIS. Logistic planning for evacuation activities and dispatching of rescue teams were improved.

THE ASSESSMENT OF CURVED CENTERLINE GENERATION IN HDMAPS BASED ON POINT CLOUDS

Abstract. Over the decades, autonomous driving technology has attracted a lot of attention and is under rapid development. However, it still suffers from inadequate accuracy in a certain area, such as the urban area, Global Navigation Satellite System (GNSS) hostile area, due to the multipath interference or Non-Line-of-Sight (NLOS) reception. In order to realize fully autonomous applications, High Definition Maps (HD Maps) become extra assisted information for autonomous vehicles to improve road safety in recent years. Compared with the conventional navigation maps, the accuracy requirement in HD Maps, which is 20 cm in the horizontal direction and 30 cm in 3D space, is considerably higher than the conventional one. Additionally, HD Maps consist of rich and high accurate road traffic information and road elements. For the requirement of high accuracy, conducting a Mobile Laser Scanning (MLS) system is an appropriate method to collect the geospatial data accurately and efficiently. Nowadays, digital vector maps are constructed by digitalizing manually on the collected data. However, the manual process spends a lot of manpower and is not efficient and practical for a large field. Therefore, this paper proposes to automatically construct the crucial road elements, such as road edge, lane line, and centerline, to generate the HD Maps based on point clouds collected by the MMS from the surveying company. The RMSEs in the horizontal direction of the road edge, lane line, and centerline are all lower than 30 cm in 3D space.

AUTOMATIC EXTRACTION OF ROAD CENTERLINES AND EDGE LINES FROM AERIAL IMAGES VIA CNN-BASED REGRESSION

Abstract. Extracting roads from aerial images is a challenging task in the field of remote sensing. Most approaches formulate road extraction as a segmentation problem and use thinning and edge detection to obtain road centerlines and edge lines, which could produce spurs around the extracted centerlines/edge lines. In this study, a novel regression-based method is proposed to extract road centerlines and edge lines directly from aerial images. The method consists of three major steps. First, an end-to-end regression network based on CNN is trained to predict confidence maps for road centerlines and estimate road width. Then, after the CNN predicts the confidence map, non-maximum suppression and road tracking are applied to extract accurate road centerlines and construct road topology. Meanwhile, Road edge lines are generated based on the road width estimated by the CNN. Finally, in order to improve the connectivity of extracted road network, tensor voting is applied to detect road intersections and the detected intersections are used as guidance for the overcome of discontinuities. The experiments conducted on the SpaceNet and DeepGlobe datasets show that our approach achieves better performance than other methods.

Applying deep neural networks for multi-level classification of driver drowsiness using Vehicle-based measures

Drowsy driving is one of the main causes of road accidents. Accurate and reliable detection of drivers' drowsiness is significantly important to prevent drowsiness-related accidents. In the context of automated vehicle driving, it is important for intelligent systems to know the current state of the driver to prepare handover maneuvers. Previous studies are mostly based on manually extracted features from either driving performance or driver physiological data. This methodology of a priori defined features can lead to losing valuable information of input signals that are significant to classify drowsiness levels in individual drivers because generally, it is not known which features are suitable for drowsiness prediction before classification. By using deep neural networks, features can be extracted automatically from preprocessed data. This paper presents a new non-obtrusive drowsiness detection system based on deep neural networks using vehicle-based measures. The proposed method is based on a combination of convolutional neural networks (CNN) and recurrent neural networks (RNN). Five vehicle-based measures, including lateral deviation from road centerline, lateral acceleration, yaw rate, steering wheel angle, and steering wheel velocity, are exploited as network inputs. The level of drowsiness is classified into three different classes. Long-short term memory (LSTM) and gated recurrent unit (GRU) layers are used as RNN in the structure of the designed deep network. The performance of the proposed method is evaluated on experimental data that were collected from 44 sessions in a fixed-base driving simulator simulating monotonous night-time highway drives. Results show that the classification accuracy of the designed deep networks outperforms traditional classifiers like support vector machine and k-nearest neighbors. The highest accuracy of 96.0% has been achieved with a combination of CNN and LSTM (CNN-LSTM). Further research should include more signal sources, including unobtrusively taken physiological signals, and test the system in real-world conditions.

Lane Keeping Control Based on Model Predictive Control Under Region of Interest Prediction Considering Vehicle Motion States

To address the failure to consider vehicle states in region of interest (ROI) prediction, we propose the use of a Kalman filter to estimate the position of vehicles relative to lanes by vehicle states on the basis of a vehicle–road micro traffic model in the world coordinate system. The central position of the ROI is determined through a combination of optimal preview time theory with the ROI prediction. The range of the ROI is determined by offsetting upward, downward, leftward, and rightward from the central position of the ROI. The left and right ROI are processed separately to detect lane lines. Simulation results show that the proposed prediction method reduces the ROI range, and the model predictive control controller can make the vehicle run smoothly from the initial position to the road centerline.

Simultaneous Road Surface and Centerline Extraction From Large-Scale Remote Sensing Images Using CNN-Based Segmentation and Tracing

Accurate and up-to-date road maps are of great importance in a wide range of applications. Unfortunately, automatic road extraction from high-resolution remote sensing images remains challenging due to the occlusion of trees and buildings, discriminability of roads, and complex backgrounds. To address these problems, especially road connectivity and completeness, in this article, we introduce a novel deep learning-based multistage framework to accurately extract the road surface and road centerline simultaneously. Our framework consists of three steps: boosting segmentation, multiple starting points tracing, and fusion. The initial road surface segmentation is achieved with a fully convolutional network (FCN), after which another lighter FCN is applied several times to boost the accuracy and connectivity of the initial segmentation. In the multiple starting points tracing step, the starting points are automatically generated by extracting the road intersections of the segmentation results, which then are utilized to track consecutive and complete road networks through an iterative search strategy embedded in a convolutional neural network (CNN). The fusion step aggregates the semantic and topological information of road networks by combining the segmentation and tracing results to produce the final and refined road segmentation and centerline maps. We evaluated our method utilizing three data sets covering various road situations in more than 40 cities around the world. The results demonstrate the superior performance of our proposed framework. Specifically, our method’s performance exceeded the other methods by 7% and 40% for the connectivity indicator for road surface segmentation and for the completeness indicator for centerline extraction, respectively.

Learning to Generate Maps from Trajectories

Accurate and updated road network data is vital in many urban applications, such as car-sharing, and logistics. The traditional approach to identifying the road network, i.e., field survey, requires a significant amount of time and effort. With the wide usage of GPS embedded devices, a huge amount of trajectory data has been generated by different types of mobile objects, which provides a new opportunity to extract the underlying road network. However, the existing trajectory-based map recovery approaches require many empirical parameters and do not utilize the prior knowledge in existing maps, which over-simplifies or over-complicates the reconstructed road network. To this end, we propose a deep learning-based map generation framework, i.e., DeepMG, which learns the structure of the existing road network to overcome the noisy GPS positions. More specifically, DeepMG extracts features from trajectories in both spatial view and transition view and uses a convolutional deep neural network T2RNet to infer road centerlines. After that, a trajectory-based post-processing algorithm is proposed to refine the topological connectivity of the recovered map. Extensive experiments on two real-world trajectory datasets confirm that DeepMG significantly outperforms the state-of-the-art methods.

Automatic Seamline Determination for Urban Image Mosaicking Based on Road Probability Map from the D-LinkNet Neural Network.

Image mosaicking which is a process of constructing multiple orthoimages into a single seamless composite orthoimage, is one of the key steps for the production of large-scale digital orthophoto maps (DOM). Seamline determination is one of the most difficult technologies in the automatic mosaicking of orthoimages. The seamlines that follow the centerlines of roads where no significant differences exist are beneficial to improve the quality of image mosaicking. Based on this idea, this paper proposes a novel method of seamline determination based on road probability map from the D-LinkNet neural network for urban image mosaicking. This method optimizes the seamlines at both the semantic and pixel level as follows. First, the road probability map is obtained with the D-LinkNet neural network and related post processing. Second, the preferred road areas (PRAs) are determined by binarizing the road probability map of the overlapping area in the left and right image. The PRAs are the priority areas in which the seamlines cross. Finally, the final seamlines are determined by Dijkstra’s shortest path algorithm implemented with binary min-heap at the pixel level. The experimental results of three group data sets show the advantages of the proposed method. Compared with two previous methods, the seamlines obtained by the proposed method pass through the less obvious objects and mainly follow the roads. In terms of the computational efficiency, the proposed method also has a high efficiency.

RESEARCH AND IMPLEMENTATION OF 3D CITY BATCH RAPID MODELLING METHOD

Abstract. For the current 3D city modelling, the use of computer programs for batch modelling of non-fine models can only ensure the building height information, and cannot effectively use the attribute information of 2D data; and the manual modelling method for the road ancillary facilities has the disadvantages of modelling and placement of streetlamps, low efficiency, and inaccurate positions, a 3D modelling method based on parametric modelling technology and 3DMax modelling technology to realize batch modelling of non-fine buildings and urban streetlamps was proposed. Firstly, the spatial and attribute information of two-dimensional surveying data is used as basic data; then, through the geometric functions provided by CityEngine, combined with the characteristics of the main building and roof structure of the village in the city, the functions of generating the building structure is constructed, and the batch-based automatic modelling is compiled. The program uses the bottom attribute information to control the structure and texture of the model. . The external model was introduced through geometric function, use the attribute information and the adjacent angle of the road centerline to control the style, size and direction of the streetlamp, realize the batch automatic construction at the streetlamp coordinate point. Finally, the Python language bulk export model plug-in and a MaxScript script bulk import model plugin are compiled to improve work efficiency and model compatibility. Through experiments and performance analysis, it is shown that the method can guarantee the rapid establishment of 3D models of buildings and streetlamps, and the structure and texture are vivid. It is well compatible with 3DMax and can be directly modified and format converted.

Semiautomatic Road Extraction Framework Based on Shape Features and LS-SVM from High-Resolution Images

Road network delineation from remote sensing imagery is useful in many applications related to humans in present days as it provides meaningful information which is very helpful in the field of transportation and other support systems of human civilization. However, the road network extraction from remote sensing images is a challenging problem because of various properties of road network altered by different factors. In this paper, a supervised multistage framework based on least square support vector machine (LS-SVM), mathematical morphology and road shape features is proposed to extract road networks from remote sensing images. In the first stage, after denoising the image using the nonlinear filter, the image is segmented using LS-SVM into two classes’ road or non-road but some non-road components like building structures, shadows of trees and other roads like objects are also classified as the road network. In the second stage, the morphology and primary shape features are used to remove the non-road objects. Finally, road centerlines are extracted using a method based on Euclidean distance transformation. The experimental results reflect the accuracy of the proposed work as compared to other road extraction methods.

A Self-Supervised Learning Framework for Road Centerline Extraction From High-Resolution Remote Sensing Images

Road extraction from the high-resolution remote sensing image is significant for the land planning, vehicle navigation, etc. The existing road extraction methods normally need many preprocessing and subsequent optimization steps. Therefore, an automatic road centerline extraction method based on the self-supervised learning framework for high-resolution remote sensing image is proposed. This proposed method does not need to manually select training samples and other optimization steps, such as the nonroad area removing. First, the positive sample selection method combining the spectral and shape features is proposed to extract the road sample. Then, the one-class classifier framework is introduced and the random forest positive unlabeled learning classifier is constructed to get the posterior probability of the pixel belonging to road. The shape feature and the posterior probability are combined to form the final road network in the object-oriented way. Finally, the road centerline is obtained through the tensor voting algorithm. In order to verify the effectiveness of the proposed algorithm, high-resolution remote sensing images and benchmark datasets are used to do experiments. The indexes of the completeness ratio, the correctness ratio, and the detection quality are used for the quantitative accuracy evaluation. Compared with the supervised, the unsupervised, and the one-class classification road extraction algorithms, this proposed algorithm achieves high accuracy and efficiency. For the deep learning method comparison, the deep learning method performs well in most cases especially in the complex urban area. However, the deep learning method needs a large number of samples and a long training time, and our self-supervised learning framework does not need the training samples.

Density Adaptive Approach for Generating Road Network From GPS Trajectories

Road networks are fundamental parts of intelligent transportation and smart cities. With the emergence of crowdsourcing geographic data, road mapping approaches by using crowdsourcing trajectories have been developed. Existing road map inference algorithms from trajectories can extract relatively accurate road networks, however, these algorithms are not robust to different trajectory datasets and the parameter optimization task is tedious and time-consuming. Therefore, we propose an adaptive approach based on trajectory density. The proposed approach contains two stages. Firstly, the density distribution for each trajectory is adaptively estimated by the Gaussian fitting approach and the density peak points are extracted to construct road centerlines corresponding to each trajectory. Secondly, these extracted road centerlines are incrementally merged by the “matching-refinement-merging” process to generate a road network. We compare the proposed approach against four representative methods through trajectory datasets that are completely different in sampling frequency, trajectory density, road density, and noise. The results show that the proposed approach provides better accuracy in terms of precision and integrity and does not require additional parameter adjustment.

DeepWindow: Sliding Window Based on Deep Learning for Road Extraction From Remote Sensing Images

The road centerline extraction is the key step of the road network extraction and modeling. The hand-craft feature engineering in the traditional road extraction methods is unstable, which makes the extracted road centerline deviated from the road center in complex cases and even results in overall extracting errors. Recently, the road centerline extraction methods based on semantic segmentation employing deep neural network greatly outperformed the traditional methods. Nevertheless, the pixel-wise labels for training deep learning models are expensive and the postprocess of road segmentation is error-prone. Inspired by the work of human pose estimation, we propose DeepWindow, a novel method to automatically extract the road network from remote sensing images. DeepWindow uses a sliding window guided by a CNN-based decision function to track the road network directly from the images without the prior of road segmentation. First of all, we design and train a CNN model to estimate the road center points inside a patch. Then, the road seeds are automatically searched patch by patch employing the CNN model. Finally, starting from seeds, our method first estimates the road direction using a Fourier spectrum analysis algorithm and then iteratively tracks the road center-line along the road direction guided by the CNN model. In our method, the CNN model is trained by point annotations, which greatly reduces the training costs comparing to those in semantic model training. Our method achieves comparable performance with the state-of-the-art road extraction methods, and extensive experiments indicate that our method is robust to the point deviation.

Road centreline and lane reconstruction from pervasive GPS tracking on motorways

Abstract A lane-based Routable Digital Map is a basis for construction of a floating-car dataset for lane-based traffic analysis. We proposed an algorithm that is capable of identifying lanes in highway segments based on GPS trajectories collected by mobile phones. The algorithm consists of three main steps. First, we identify nodes within the test site, and divide the network into segments. Second, the central line of each segment is identified based on a dissimilarity matrix computed based on Dynamic Time Warping criteria. Finally, the Gaussian Mixture Method is used to identify the lanes. This allows the width of the lanes to remain constant throughout the segment. The results have been validated by comparing the share of traffic volume in each lane based on the trajectory points in the identified lanes and the loop detectors’ data. The results show that the proposed algorithm can determine the lanes with acceptable accuracy. Estimating the traffic volume and its speed based on floating car data provides a big step in enabling data fusion from multiple sources more accurately and to estimate traffic state more precisely.

USING OF VHR SATELLITE IMAGES FOR ROAD NETWORK EXTRACTION IN EGYPT

Roads extraction from VHR satellite images are very paramount for GIS and map updating. Due to the high resolution of satellite images, there are many obstacles broken roads such as shadow, and vehicles. The present work aims to find the most suitable road extraction approach that can be applied in the Egyptian environment. In this study, two satellite images from WorldView-2 and WorldView-3 were used. Classification of image by pixel-based and object-based was carried out to find the appropriate classification method for road extraction. Then, road class refinement by morphology and angular texture signature are performed to decrease the misclassifications between roads and other spectrally similar objects. After that, an iterative and localized Hough transform method was compared with the thinning algorithm method to find the proper method that can extract road centerline segments from the refined images. The performance of the extracted roads was estimated by using the common metrics; completeness, correctness, and quality. The results of this work demonstrate that the random tree in object-based classification achieves the highest overall accuracy than other classification methods. Also, thinning algorithm has more advantages than Hough transform.

Research on Evolutionary Design Method of BIM Technology in Highway Reconstruction and Expansion Project

BIM technology is widely used in expressway Construction projects. In the process of transforming two-dimensional design into three-dimensional design, there are still many problems that need to be solved urgently. Based on the Beijing-Shanghai highway reconstruction and expansion project, this paper proposes an evolutionary design method. Based on the application standard of project model technology, the method uses BIM model software, based on skeleton lines and templates, using EKL programming language to realize the structure size from coarse to fine, information from summary to detail, and progressive design method for component granularity from bottom to high. At the same time, a three-dimensional design method for automatically generating road centerlines is proposed. The results show that the evolutionary design method is in line with the current project habits, matching the existing design concepts, and can efficiently complete the design work at different stages, effectively avoiding a lot of repetitive work.

A Unifying Theory of Driver Perception and Steering Control on Straight and Winding Roads

Novel driver support systems potentially enhance road safety by cooperating with the human driver. To optimize the design of emerging steering support systems, a profound understanding of driver steering behavior is required. This article proposes a new theory of driver steering, which unifies visual perception and control models. The theory is derived directly from measured steering data, without any a priori assumptions on driver inputs or control dynamics. Results of a human-in-the-loop simulator experiment are presented, in which drivers tracked the centerline of straight and winding roads. Multiloop frequency response function (FRF) estimates reveal how drivers use visual preview, lateral position feedback, and heading feedback for control. Classical control theory is used to model all three FRF estimates. The model has physically interpretable parameters, which indicate that drivers minimize the bearing angle to an “aim point” (located 0.25–0.75 s ahead) through simple compensatory control, both on straight and winding roads. The resulting unifying perception and control theory provides a new tool for rationalizing driver steering behavior, and for optimizing modern steering support systems.

Multi-Scale and Multi-Task Deep Learning Framework for Automatic Road Extraction

Road detection and centerline extraction from very high-resolution (VHR) remote sensing imagery are of great significance in various practical applications. Road detection and centerline extraction operations depend on each other, to a certain extent. The road detection constrains the appearance of the centerline, and the centerline enhances the linear features of the road detection. However, most of the previous works have addressed these two tasks separately and have not considered the symbiotic relationship between them, making it difficult to obtain smooth and complete roads. In this paper, a novel multi-scale and multi-task deep learning framework for automatic road extraction (MSMT-RE) is proposed to build the relationship between them and simultaneously complete the road detection and centerline extraction tasks. U-Net is selected as the basic network for multi-task learning due to its strong ability to preserve spatial details. Multi-scale feature integration is also applied in the framework to increase the robustness of the feature extraction. Meanwhile, an adaptive loss function is introduced to solve the problems of roads taking up a small percentage of the training samples, and the fact that the positive samples of the two tasks are unbalanced. Finally, experiments were conducted on two public road data sets and two large images from Google Earth, and the proposed framework was compared with other state-of-the-art deep learning-based road extraction methods, both quantitatively and qualitatively. The proposed approach outperformed all the compared methods, confirming its advantages in automatic road extraction.

Road Extraction from Very High Resolution Images Using Weakly labeled OpenStreetMap Centerline

Road networks play a significant role in modern city management. It is necessary to continually extract current road structure, as it changes rapidly with the development of the city. Due to the success of semantic segmentation based on deep learning in the application of computer vision, extracting road networks from VHR (Very High Resolution) imagery becomes a method of updating geographic databases. The major shortcoming of deep learning methods for road networks extraction is that they need a massive amount of high quality pixel-wise training datasets, which is hard to obtain. Meanwhile, a large amount of different types of VGI (volunteer geographic information) data including road centerline has been accumulated in the past few decades. However, most road centerlines in VGI data lack precise width information and, therefore, cannot be directly applied to conventional supervised deep learning models. In this paper, we propose a novel weakly supervised method to extract road networks from VHR images using only the OSM (OpenStreetMap) road centerline as training data instead of high quality pixel-wise road width label. Large amounts of paired Google Earth images and OSM data are used to validate the approach. The results show that the proposed method can extract road networks from the VHR images both accurately and effectively without using pixel-wise road training data.

Evaluating rider steering responses to an unexpected collision hazard using a motorcycle riding simulator

Motorcycle rider steer responses to unexpected collision emergencies have not been studied experimentally. We used a motorcycle simulator with elastic steer mechanism and modified car driving model to simulate the input-output counter steering response of two-wheeled vehicles in combination with a car pop-up paradigm from driving studies to evaluate rider responses to unexpected collision hazards. We manipulated time-to-crash – either 1 s or 1.5 s – to probe the threshold between not enough and just enough time to respond. The median response time of 570 ms was similar to previously recorded latencies for motorcycle emergency braking responses. Although median response time and steer torque did not depend on time-to-crash (TTC), the distribution and variability of response measures were increased when TTC was shorter. Riders showed improved lateral displacement toward the road centerline by the third TTC1.5 trial and were almost 8 times more likely to produce a successful virtual swerve avoidance maneuver in TTC1.5 rather than TTC1 trials. With 1.5 s to respond, riders were more consistent, with net steering inputs more congruent with the maneuvering goal compared to when they had only 1 s to respond. Comparisons of ‘safe’ versus ‘crash’ outcomes show that average response times across trials and riders are not different, but the variance in timing is lower across successful trials. We have shown than it is possible to safely study rider reactions to an emergency and observe a realistic range of steer responses to a traffic conflict using a motorcycle riding simulator. The results have relevance for the design of automatic rider assistive systems, rider behavior prediction and decision logic algorithms, PTW rider modeling and targeted training interventions.