Search

This research focuses on improving the modelling of highly automated centralized traffic management. Authors use a binary integer modelling for traffic flow optimization. One of the main results of the research is the integration of alternative graph structures simultaneously in the investigated traffic optimization model. This allows to take into account different categories of vehicles, especially large vehicles, and specific traffic directions. The study highlights the need for seamless interoperability between graph structures and demonstrates its feasibility through the investigation of advanced safety indicators. In addition, methods are developed and presented that will allow the future integration of non-automated traffic elements and conventional traffic engineering solutions. During our research, we reviewed the automated traffic management models, focusing on the advantages of both continuous and discrete models. Continuous models provide realism but are computationally intensive, while discrete models are simpler but less realistic. Considering this, our paper proposes solutions that offer a trade-off between these approaches, allowing a balance between computational complexity, accuracy, and realism in vehicle modelling. [ABSTRACT FROM AUTHOR]

This research focuses on improving the modelling of highly automated centralized traffic management. Authors use a binary integer modelling for traffic flow optimization. One of the main results of the research is the integration of alternative graph structures simultaneously in the investigated traffic optimization model. This allows to take into account different categories of vehicles, especially large vehicles, and specific traffic directions. The study highlights the need for seamless interoperability between graph structures and demonstrates its feasibility through the investigation of advanced safety indicators. In addition, methods are developed and presented that will allow the future integration of non-automated traffic elements and conventional traffic engineering solutions. During our research, we reviewed the automated traffic management models, focusing on the advantages of both continuous and discrete models. Continuous models provide realism but are computationally intensive, while discrete models are simpler but less realistic. Considering this, our paper proposes solutions that offer a trade-off between these approaches, allowing a balance between computational complexity, accuracy, and realism in vehicle modelling.

Autonomous navigation of unmanned aerial vehicles (UAVs) has become in the past decade an extremely attracting topic, also due to the increasing availability of affordable equipment and open-source control and processing software environments. This demand has also raised a strong interest in developing accessible experimental platforms to train engineering students in the rapidly evolving area of autonomous navigation. In this paper, we describe a platform based on low-cost off-the-shelf hardware that takes advantage of the Matlab/Simulink programming environment to tackle most of the problems related to UAV autonomous navigation. More specifically, the subject of this paper is the autonomous control of the flight of a small UAV, which must explore and patrol an indoor unknown environment. Objectives: to analyse the existing hardware platforms for autonomous flight indoors, choose a flight exploration scenario of unknown premises, to formalize the procedure for obtaining a model of knowledge for semantic classification of premises, to formalize obtaining distance to obstacles using data camera horizontally employment and building on its barrier map. Namely, we use the method of image segmentation based on the brightness threshold, a method of training the semantic segmentation network, and computer algorithms in probabilistic robotics for mobile robots. We consider both the case of navigation guided by structural visual information placed in the environment, e.g., contrast markers for flight (such as path marked by a red tape), and the case of navigation based on unstructured information such as recognizable objects or human gestures. Basing on preliminary tests, the most suitable method for autonomous in-door navigation is by using\ object classification and segmentation, so that the UAV gradually analyses the surrounding objects in the room and makes decisions on path planning. The result of our investigation is a method that is suitable to allow the autonomous flight of a UAV with a frontal video camera. Conclusions. The scientific novelty of the obtained results is as follows: we have improved the method of autonomous flight of small UAVs by using the semantic network model and determining the purpose of flight only at a given altitude to minimize the computational costs of limited autopilot capabilities for low-cost small UAV models. The results of our study can be further extended by means of a campaign of experiments in different environments.

It is important that indoor autonomous mobile platforms have the capability of localization in general indoor environments. In this study, using multi-threaded vehicle-mounted light detection and ranging (LiDAR), we conducted indoor autonomous mobile platform localization experiments based on a point cloud conversion 2D image method with an interpolated probability distribution, performed a scan matching analysis by converting 2D images based on an interpolated probability distribution while using occupied grid maps, and introduced a multiresolution map method to avoid falling into a local optimum. We found that the method adopted in this study achieves a higher indoor positioning accuracy and a higher matching speed with reduced computational effort while avoiding local optima. Compared with other traditional indoor positioning methods, this method has the advantages of universal applicability and robustness against signal interference and other problems. [ABSTRACT FROM AUTHOR]

The detection of dynamic and static obstacles is a key task for the navigation of autonomous ground vehicles. The article presents a new algorithm for generating an occupancy map of the surrounding space from noisy point clouds obtained from one or several stereo cameras. The camera images are segmented by the proposed deep neural network FCN-ResNet-M-OC, which combines the speed of the FCN-ResNet method and improves the quality of the model using the concept of object context representation. The paper investigates supervised approaches to network training on unbalanced samples with road scenes such as the weighted cross entropy and the Focal Loss. The occupancy map is built from point clouds with semantic labels, in which static environment and potentially dynamic obstacles are highlighted. Our solution is operational in real time and applicable on platforms with limited computing resources. The approach was tested on autonomous vehicle datasets: Semantic KITTI, KITTI-360, Mapillary Vistas and custom OpenTaganrog. The usage of semantically labeled point clouds increased the precision of obstacle detection by an average of 17%. The performance of the entire approach on various computing platforms with Jetson Xavier, RTX3070, GPUs NVidia Tesla V100 is respectively from 10 to 15 FPS for input image resolution $1920\times 1080$ pixels. [ABSTRACT FROM AUTHOR]

To enable autonomous operation for Maritime Autonomous Surface Ships (MASS), a reliable environment sensing algorithm is required. This is because MASS may operate in complex waters with large numbers of static and dynamic objects, and environment data acquired from either onboard sensors or external information providing systems is subject to both random errors and reliability issue in harsh weather. Many existing marine radar target tracking algorithms implicitly assume the environment only consists of dynamic objects and ignore the existence of static objects with various structure and size, limiting their ability in utilizing knowledge from third parties such as electronic navigation chart. In this paper, an occupancy grid model (OGM) is constructed as a fundamental environment sensing framework. Mapping sensor data to occupancy grids can facilitate identification and inference on static objects as well as dynamic targets. Hidden Markov models (HMMs) are then introduced for tracking of moving vessels among discrete grids. The proposed algorithm employs a hierarchical structure to make use of both ENC and marine radar data for environment sensing. Experiment validation based on both simulation and real data shows that it is capable of identifying static objects and simultaneously tracking moving vessels using occupancy grids. • An occupancy grid based environment model as a fundamental sensing framework for MASS. • An HMM-based algorithm for vessel target tracking in grids. • Discussion on the benefits of occupancy grid formulation. • Improvement in consistent environment sensing for MASS. [ABSTRACT FROM AUTHOR]

Radar sensors have great advantages over other sensors in estimating the motion states of moving objects, because they detect velocity components within one measurement cycle. Moreover, numerous successes have already been achieved regarding the classification of such objects. However, the advantage of instantaneous velocity measurement is lost when detecting static objects, so that their classification is much more demanding. In this paper, we use semantic segmentation networks to distinguish between frequently occurring infrastructure objects. The resulting semantic grids provide a location-based classification of the vehicle environment. Since even modern radars have a significantly poorer angular resolution than lidars, the relatively thin radar point cloud is accumulated in advance and transformed into 2D or 3D grids that act as network inputs. Occupancy grids are particularly advantageous here, since they calculate not only the obstacles but also the free spaces. With suitable parameter selection, which is very challenging due to the complexity of radar measurement, the resulting grids allow for good association with camera images. Finally, in order to evaluate possible advantages of 3D grids as network input with respect to the segmentation result, we created and evaluated a simulation dataset and two different real-world datasets in car parks and on motorways. As a result, Jaccard coefficients between 81% and 88% were achieved, depending on the dataset. It was also found that 3D input images lead to improvements in the car park dataset.

In this paper, we propose a lane change decision scheme using only commercial, automotive radar sensors with "DRiving Environment Assessment and decision Making (DREAM) index." First, the index associated with a risk situation is assessed by the concept of a dynamic occupancy grid zone. The predefined distance of the local area was set to conform to the international standard of a steering function. Also, the risk assessment was predictively conducted using modeling of the relative motion with a target vehicle. Second, the index associated with a cooperative driving concept of the surrounding vehicles is proposed. To estimate the relative acceleration which is not directly measured by radar, we designed a discrete‐time state estimator. We executed scenario‐based experiments with test vehicles in a high‐speed circuit to validate the decision scheme. Through the experiments, we observed that the DREAM index could make effective decisions, and the lane change maneuverings were performed successfully in real‐world tests. [ABSTRACT FROM AUTHOR]

This study presents the process employed in prototyping and early evaluation of automotive perception algorithms. The data generation was performed using an automotive virtual validation tool. The off-the-shelf simulation framework used was expanded to include phenomenological sensors model that allowed for a simplified simulation of radars, lidars, and cameras. This paper extends the description of the methods for the generation of control algorithms. The work presented also includes a description of relevant data fusion methods for building occupancy grids. Results were obtained by performing a comparison of algorithm results against ground-truth. This virtual validation was used to enable early definition and verification of system-level requirements, narrow down performance assessment methods, and identify performance limitations before data from real sensors are available. [ABSTRACT FROM AUTHOR]

With the development of RGB-D cameras, dense point cloud model gains great attention for its information richness and obstacle avoidance features. It can be overlapped to occupancy grid for path planning and navigation applications. But there are redundant information since point cloud models tend to perceive every details in the environment and the computation complexity of traversal increases significantly with scene expansion. A possible solution is the combination of measurements of different granularities from various sensors to construct the environment models in uniform representation. Based on octree occupancy grid and our previous work, we propose a multi-granularity environment perception algorithm, which uniformly represents environment models from various sensors. A probabilistic octree representation is constructed to uniformly express the point cloud models. This representation uniformly fuses the sparse, semi-dense and dense models dynamically through an incremental algorithm along with the camera trajectory. Multiple resolutions of the same model can be obtained at any time by limiting the depth of a query. Experiments demonstrate the effectiveness of our method in minimizing trajectory error on several public available benchmarks and reducing the space complexity of environment models. [ABSTRACT FROM AUTHOR]

In the context of autonomous robots, one of the most important tasks is to prevent potential damage to the robot during navigation. For this purpose, it is often assumed that one must deal with known probabilistic obstacles, then compute the probability of collision with each obstacle. However, in complex scenarios or unstructured environments, it might be difficult to detect such obstacles. In these cases, a metric map is used, where each position stores the information of occupancy. The most common type of metric map is the Bayesian occupancy map. However, this type of map is not well suited for computing risk assessments for continuous paths due to its discrete nature. Hence, we introduce a novel type of map called the Lambda Field, which is specially designed for risk assessment. We first propose a way to compute such a map and the expectation of a generic risk over a path. Then, we demonstrate the benefits of our generic formulation with a use case defining the risk as the expected collision force over a path. Using this risk definition and the Lambda Field, we show that our framework is capable of doing classical path planning while having a physical-based metric. Furthermore, the Lambda Field gives a natural way to deal with unstructured environments, such as tall grass. Where standard environment representations would always generate trajectories going around such obstacles, our framework allows the robot to go through the grass while being aware of the risk taken.

Several problems can be encountered in the design of autonomous vehicles. Their software is organized into three main layers: perception, planning, and actuation. The planning layer deals with the sort and long-term situation prediction, which are crucial for intelligent vehicles. Whatever method is used to make forecasts, vehicles’ dynamic environment must be processed for accurate long-term forecasting. In the present article, a method is proposed to preprocess the dynamic environment in a freeway traffic situation. The method uses the structured data of surrounding vehicles and transforms it to an occupancy grid which a Convolutional Variational Autoencoder (CVAE) processes. The grids (2048 pixels) are compressed to a 64-dimensional latent vector by the encoder and reconstructed by the decoder. The output pixel intensities are interpreted as probabilities of the corresponding field is occupied by a vehicle. This method’s benefit is to preprocess the structured data of the dynamic environment and represent it in a lower-dimensional vector that can be used in any further tasks built on it. This representation is not handmade or heuristic but extracted from the database patterns in an unsupervised way.

Representing the environment is a fundamental task in enabling robots to act autonomously in unknown environments. In this work, we present confidence-rich mapping (CRM), a new algorithm for spatial grid-based mapping of the 3D environment. CRM augments the occupancy level at each voxel by its confidence value. By explicitly storing and evolving confidence values using the CRM filter, CRM extends traditional grid mapping in three ways: first, it partially maintains the probabilistic dependence among voxels; second, it relaxes the need for hand-engineering an inverse sensor model and proposes the concept of sensor cause model that can be derived in a principled manner from the forward sensor model; third, and most importantly, it provides consistent confidence values over the occupancy estimation that can be reliably used in collision risk evaluation and motion planning. CRM runs online and enables mapping environments where voxels might be partially occupied. We demonstrate the performance of the method on various datasets and environments in simulation and on physical systems. We show in real-world experiments that, in addition to achieving maps that are more accurate than traditional methods, the proposed filtering scheme demonstrates a much higher level of consistency between its error and the reported confidence, hence, enabling a more reliable collision risk evaluation for motion planning. [ABSTRACT FROM AUTHOR]

This paper proposes a method that utilizes a 3-D occupancy grid to efficiently map a large area while retaining simple representations of objects for path planning and provide spatial characteristics of objects, which may be used for object classification. To enable large-scale mapping of objects, a region around the unmanned surface vehicle (USV) is defined where a high density of LiDAR returns is expected, termed the visibility horizon. The polygon intersection between the visibility horizon and the newly detected objects is computed, as well as the polygon subtraction of the visibility horizon from the mapped list of polygons. The two polygon lists are then combined using a polygon union operation, with the objects retaining class designations. The result is a 2-D map that contains polygon representations of objects, where the object is described with a tunable number of vertices and may have an associated object class. Thus, providing necessary information for path planning and tasking. The resultant polygons are shown here to be accurate to 20 cm using a 10-cm occupancy grid and 16-ft-long unmanned surface vehicles with four multibeam LiDAR sensors. [ABSTRACT FROM AUTHOR]

Copyright of Journal of Chongqing University of Posts & Telecommunications (Natural Science Edition) is the property of Chongqing University of Posts & Telecommunications and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Sonar sensor based range measurement for mobile robot mapping/perception of the complex environment is considered as proficient and low cost technique. Sonar sensor are employ for various applications such as autonomous navigation of mobile robot, underwater navigation and mapping, perception of shape and size of objects, etc. Different configurations are used in the past to mount the sonar sensors on the front/back/side of the mobile robots for the accurate and precise perception of the environment. The fast, precise and consistent sonar occupancy grid mapping is the key requirement for the various mobile robot applications which results in the use of multiple sonar sensors in the form of convex shape sonar rings mounted on mobile robot. Sonar rings provide fast mapping of the environment but, at the cost of low reliability, less precision and inconsistency in the occupancy grid due to specular reflection, false triggering synchronization and cross talk. This paper presents a novel technique for the construction of concave shape sonar rings called C-type arrangement (CTA) of sonar sensors which is derived from the analytical and mathematical study of convex type sonar sensor rings, circle geometry, trigonometry and Sonar sensor model. CTA approach is designed with an objective of examine metrological characteristic of multi sonar rings such as uncertainty, precision and error in range measurement. CTA technique reduce the uncertainties in sonar occupancy grid mapping due to specular reflection and cross talk for the application such as autonomous navigation, path planning, mapping, localization, etc. The proposed technique is experimentally verified in various real world experiments and enhanced results are analyzed both quantitatively and qualitatively that validate the reliability of the CTA technique. The CTA technique enhance the reliability in sonar sensor range measurement for the various robotics task like space exploration, mining, search and rescue operation, underwater surveillance etc. [ABSTRACT FROM AUTHOR]

The quality of environmental perception is crucial for automated vehicle capabilities. In order to ensure the required accuracy, the occupancy grid mapping algorithm is often utilised to fuse data from multiple sensors. This paper focuses on the radar-based occupancy grid for highway applications and describes how to measure effectively the quality of the occupancy map. The evaluation was performed using the novel grid pole-like object analysis method. The proposed assessment is versatile and can be applied without detailed ground truth information. The evaluation was tested with a simulation and real vehicle experiments on the highway.

Today, perception solutions for Automated Vehicles rely on sensors on board the vehicle, which are limited by the line of sight and occlusions caused by any other elements on the road. As an alternative, Vehicle-to-Everything (V2X) communications allow vehicles to cooperate and enhance their perception capabilities. Besides announcing its own presence and intentions, services such as Collective Perception (CPS) aim to share information about perceived objects as a high-level description. This work proposes a perception framework for fusing information from on-board sensors and data received via CPS messages (CPM). To that end, the environment is modeled using an occupancy grid where occupied, and free and uncertain space is considered. For each sensor, including V2X, independent grids are calculated from sensor measurements and uncertainties and then fused in terms of both occupancy and confidence. Moreover, the implementation of a Particle Filter allows the evolution of cell occupancy from one step to the next, allowing for object tracking. The proposed framework was validated on a set of experiments using real vehicles and infrastructure sensors for sensing static and dynamic objects. Results showed a good performance even under important uncertainties and delays, hence validating the viability of the proposed framework for Collective Perception.