Real Traffic Scenarios Research Articles

Autonomous driving planning and decision making based on game theory and reinforcement learning

AbstractAutonomous driving technology is one of the important methods that avoid the hidden dangers of traffic safety. Although the existing autonomous driving technology can meet some needs of real traffic scenarios, the ability of multi‐agents to effectively generate autonomous driving strategy in complex traffic environments remains to be improved. Aiming at this problem, the automatic drive model based on game theory and reinforcement learning is proposed by combining these two technologies and applying them in multi‐agent cooperative driving, which enables multi‐agents to carry out strategic reasoning with negotiation in traffic scenarios by extending the game description language, and puts forward the constrained multi‐agents deep deterministic policies gradient algorithm. Finally, the related experiments are conducted through the autonomous driving simulation platform, and the experimental results show that the proposed model can effectively generate driving strategies for multi‐agents in complex traffic environments, which verifies the validity and feasibility of the proposed model, and provides the general research basis for multi‐agents autonomous driving.

Audiovisual time-to-collision estimation for accelerating vehicles: The acoustic signature of electric vehicles impairs pedestrians' judgments

To avoid collisions, pedestrians intending to cross a road need to accurately estimate the time-to-collision (TTC) of an approaching vehicle. For TTC estimation, auditory information can be considered particularly relevant when the approaching vehicle accelerates. The sound of vehicles with internal combustion engine (ICEVs) provides characteristic auditory information about the acceleration state (increasing rotational speed and engine load). However, for electric vehicles (EVs), the acoustic signature during acceleration is less salient. Although the auditory detection of EVs has been studied extensively, there is no research on potential effects of the altered acoustic signature of EVs on TTC estimation. To close this gap, we compared TTC estimates for ICEVs and for EVs with and without activated acoustic vehicle alerting system (AVAS). We implemented a novel interactive audiovisual virtual-reality system for studying the human perception of approaching vehicles. Using acoustic recordings of real vehicles as source signals, the dynamic spatial sound field corresponding to a vehicle approaching in an urban setting is generated based on physical modeling of the sound propagation between vehicle and pedestrian (listener) and is presented via sound field synthesis (higher-order Ambisonics). In addition to the auditory simulations, the scene was visually presented on a head-mounted display with head tracking. Participants estimated the TTC of vehicles that either approached at a constant speed or accelerated positively. In conditions with constant speed, TTC estimates for EVs with and without AVAS were similar to those for ICEVs. In contrast, for accelerating vehicles, there was a substantial effect of the vehicle type on the TTC estimates. For the EVs, the mean TTC estimates showed a significant overestimation. Thus, subjects on average perceived the time of arrival of the EV at their position as longer than it actually was. The extent of overestimation increased with acceleration and presented TTC. This pattern is similar to a first-order TTC estimation representing a failure to consider the acceleration, which is consistently reported in the literature for visual-only presentations of accelerating objects. In comparison, the overestimation of TTC was largely reduced for the accelerating ICEVs. The AVAS somewhat improved the TTC estimates for the accelerating EVs, but without reaching the same level of accuracy as for the ICEVs. In real traffic scenarios, overestimations of the TTC of approaching vehicles might lead to risky road-crossing decisions. Therefore, our finding that pedestrians are significantly less able to use the acoustic information emitted by accelerating EVs for their TTC judgments, compared to accelerating ICEVs, has important implications for road safety and for the design of AVAS technologies.

A top–down approach for building realistic reference scenarios and simulation framework for LTE C-V2X communications

Many vehicle-to-everything (V2X) applications and use cases require their feasibility to be simulated, tested, and validated in realistic traffic scenarios and under various network conditions before real-time testbed implementation. As cellular-V2X (C-V2X) becomes a superior technology for future connected and autonomous vehicles, the need for a simulation framework, which integrates traffic and network simulators with a realistic channel model, becomes more evident. The challenge is to overcome existing simulation platforms’ weaknesses and improve simulation results’ accuracy while preserving flexibility with manageable implementation complexity. This paper proposed a top–down approach for building reference scenarios with macroscopic and microscopic layers, which interweaves traffic, network, and channel simulators. The basis for the proposed simulation framework is realistic scenario data, which provides the input to the road traffic simulator and the radio channel simulator. The core of the whole simulator is the network simulator interacting with the two other simulators via dedicated interfaces. The road traffic simulator generates the vehicles’ positions and is forwarded to the network simulator, which requests the radio channel simulator for the path loss values on the transmitter–receiver (TX-RX) link. As a proof of concept, the simulation results focused on the load and interference analysis in the absence and presence of V2X traffic.

The Riemann problem for a traffic flow model on a road with variable widths

Abstract We propose and investigate a model describing the traffic flow on a road with variable widths in this paper. The model, which is modified on the Aw–Rascle model, is not conservative because of the source term generated from the real traffic scenarios. The model also takes the stationary wave into considerations in the traffic flow based on the balance laws. We obtain the elementary waves of the new traffic flow model, including rarefaction waves, shock waves, contact discontinuities and particularly stationary waves. The Riemann problems to the system for the traffic flows are solved constructively, which can be further used to design numerical schemes and applied to more complex traffic flows in the future study. Some real traffic flow simulations are given, which verify the effectiveness and versatility of the model.

Distributed Stochastic Model Predictive Control for Human-Leading Heavy-Duty Truck Platoon

Human-leading truck platooning systems have been proposed to leverage the benefits of both human supervision and vehicle autonomy. Equipped with human guidance and autonomous technology, human-leading truck platooning systems are more versatile to handle uncertain traffic conditions than fully automated platooning systems. This paper presents a novel distributed stochastic model predictive control (DSMPC) design for a human-leading heavy-duty truck platoon. The proposed DSMPC design integrates the stochastic driver behavior model of the human-driven leader truck with a distributed formation control design for the following automated trucks in the platoon. The driver behavior of the human-driven leader truck is learned by a stochastic inverse reinforcement learning (SIRL) approach. The proposed stochastic driver behavior model aims to learn a distribution of cost function, which represents the richness and uniqueness of human driver behaviors, with a given set of driver-specific demonstrations. The distributed formation control includes a serial DSMPC with guaranteed recursive feasibility, closed-loop chance constraint satisfaction, and string stability. Simulation studies are conducted to investigate the efficacy of the proposed design under several realistic traffic scenarios. Compared to the baseline platoon control strategy (deterministic distributed model predictive control), the proposed DSMPC achieves superior controller performance in constraint violations and spacing errors.

InTAS - The Ingolstadt Traffic Scenario for SUMO

Vehicular Ad Hoc Networks (VANETs) are expected to be the next big step towards safer road transport, supporting applications to exchange information between vehicles. To develop novel applications for this area a high number of tests, considering all traffic situations, are demanded. However, it is unfeasible to reproduce these tests in real life, by the fact that any failure on the applications would cause severe impacts on transport system safety and could risk human lives. Thus, this paper presents the concept, model, and validation for InTAS, a realistic traffic scenario for Ingolstadt. InTAS’ road topology accurately represents Ingolstadt’s real road map. Elements such as buildings, bus stops, and traffic lights were added to the map. Twenty traffic lights systems were simulated according to the real program deployed on the traffic lights. Traffic demand was modeled based on the activitygen method, considering demographic data and real-traffic information. The city’s public transport system was also simulated accordingly to bus time-tables and their routes. The simulation step was implemented considering the best value for device.rerouting.probability, which was defined by evaluating InTAS’ output and real traffic data. The scenario was validated by comparing real-traffic data from 24 measurement points with InTAS’ simulation results.

Traffic monitoring for emergency vehicle using RFID

In developing countries like India population is significantly growing. As the population grows, the numbers of vehicles on the roads are also exponentially increasing, which results in increase in road accidents and traffic jam. Specifically, when an emergency vehicle such as Ambulance or Fire engine gets stuck in traffic jam, saving the human life becomes difficult. Under such circumstances, a promising system which can clear the traffic congestions especially in peak hours and thereby providing a safe path for emergency vehicles is very much essential. In the existing literature, less focus is given towards the problem of providing a clear path for emergency vehicles during traffic congestions. To solve these issues, a RFID-based system is proposed, which manages and regulates the traffic signals at junctions when the emergency vehicle approaches, by allowing the easy passage out of the traffic congestions. The proposed framework is modelled by means of an experimental setup using ARDUINO and LED displays which simulates a real time traffic scenario. The simulation results illustrate the better performance of the proposed framework in terms of detection as well as management of emergency vehicle by providing passage out of traffic congestions during peak hours.

A New Deep Model for Detecting Multiple Moving Targets in Real Traffic Scenarios: Machine Vision-Based Vehicles.

When performing multiple target detection, it is difficult to detect small and occluded targets in complex traffic scenes. To this end, an improved YOLOv4 detection method is proposed in this work. Firstly, the network structure of the original YOLOv4 is adjusted, and the 4× down-sampling feature map of the backbone network is introduced into the neck network of the YOLOv4 model to splice the feature map with 8× down-sampling to form a four-scale detection structure, which enhances the fusion of deep and shallow semantics information of the feature map to improve the detection accuracy of small targets. Then, the convolutional block attention module (CBAM) is added to the model neck network to enhance the learning ability for features in space and on channels. Lastly, the detection rate of the occluded target is improved by using the soft non-maximum suppression (Soft-NMS) algorithm based on the distance intersection over union (DIoU) to avoid deleting the bounding boxes. On the KITTI dataset, experimental evaluation is performed and the analysis results demonstrate that the proposed detection model can effectively improve the multiple target detection accuracy, and the mean average accuracy (mAP) of the improved YOLOv4 model reaches 81.23%, which is 3.18% higher than the original YOLOv4; and the computation speed of the proposed model reaches 47.32 FPS. Compared with existing popular detection models, the proposed model produces higher detection accuracy and computation speed.

DGInet: Dynamic graph and interaction-aware convolutional network for vehicle trajectory prediction

This paper investigates vehicle trajectory prediction problems in real traffic scenarios by fully harnessing the spatio-temporal dependencies between multiple vehicles. The existing GCN-based trajectory predictions are often considered in a single traffic scene without time attributes, complete interaction information, dynamic graph-based model, etc. Time and interaction aware models are more challenging than the existing ones. Despite very well does the graph-based model describe the relationship between driving vehicles, the critical problem in the traffic scene is how to deeply explore the spatio-temporal characteristics of dynamic graphs. Therefore, a novel dynamic graph and interaction-aware neural network model called as DGInet is proposed by combining a semi-global graph mechanism and an M-product based graph convolutional network, which are built into novel dual-network architecture in the entire model. The DGInet is built by exploiting the dynamic interaction in depth between driving vehicles in urban traffic scenarios, and then realized by utilizing semi-global graph convolution operations on the input data cell to capture the basic spatial interaction features of the driving scene. Meanwhile, the dynamic graph is further extracted by a novel M-product approach, in which the embedding of the model is then established along with the embedding of the semi-global network to perform the final embedding. Extensive experiments have been conducted on the two public datasets, named NGSIM and Apollo respectively, to show that our approach outperforms the existing ones with better performance and less computing time. Besides the real-world Shenzhen traffic dataset, China, is also developed to verify the effectiveness of our approach.

Approach for Cost Functions for the Use in Trade-Off Investigations Assessing the Environmental Impact of a Future Energy Efficient European Aviation

Aircraft emissions represent a relevant amount of human induced CO2. Globally, up to 2.5 per cent of such emissions stem from the aviation industry. In order to investigate the effects within the atmosphere, realistic flight profiles are necessary to provide quantitatively tangible values of emissions. The flight profiles and the according fuel consumption can be calculated by using waypoints from flight plans and Base of Aircraft Data (BADA). This paper presents an approach to refine the fuel consumption by integrating the passenger load into the calculation. Since effects of emissions have to be assessed on a greater scale, such as on the European air traffic network, the presented approach provides cost functions for CO2 emissions for different aircraft types and load factors. The cost functions were derived by means of regression analyses of BADA based calculated flight profiles with a step size of one second. The calculations are based on real historic traffic scenarios over several days. The derived aircraft specific fuel burn coefficients enable a simple and efficient integration of CO2 estimations depending on the flight distance, load factor and aircraft type. This can be applied to large traffic scenarios to also study different set-ups such as travel restrictions, other disruptions or an alteration in the traffic system as a whole. In order to enable the assessment of further aspects of such changes to the European air traffic system at large and to foster reproducibility and comparability of related studies, we provide further general-purpose cost estimation functions for several important key characteristics. Besides fuel consumption, we develop cost estimations for air navigation fees and maintenance for conventional aircraft. Those functions are also provided for the design concept of a short-range all-electric aircraft. This propeller aircraft features game-changing technologies such as active laminar flow control, active load alleviation and advanced materials and structure concepts. The approaches discussed in this paper will focus on the generic aspects of aircraft related costs, which can be derived from general available data. For the sake of reproducibility, the results will be made publicly available.

A hierarchical RCNN for vehicle and vehicle license plate detection and recognition

<span>Vehicle and vehicle license detection obtained incredible achievements during recent years that are also popularly used in real traffic scenarios, such as intelligent traffic monitoring systems, auto parking systems, and vehicle services. Computer vision attracted much attention in vehicle and vehicle license detection, benefit from image processing and machine learning technologies. However, the existing methods still have some issues with vehicle and vehicle license plate recognition, especially in a complex environment. In this paper, we propose a multivehicle detection and license plate recognition system based on a hierarchical region convolutional neural network (RCNN). Firstly, a higher level of RCNN is employed to extract vehicles from the original images or video frames. Secondly, the regions of the detected vehicles are input to a lower level (smaller) RCNN to detect the license plate. Thirdly, the detected license plate is split into single numbers. Finally, the individual numbers are recognized by an even smaller RCNN. The experiments on the real traffic database validated the proposed method. Compared with the commonly used all-in-one deep learning structure, the proposed hierarchical method deals with the license plate recognition task in multiple levels for sub-tasks, which enables the modification of network size and structure according to the complexity of sub-tasks. Therefore, the computation load is reduced.</span>

Action-ViT: Pedestrian Intent Prediction in Traffic Scenes

Pedestrian crossing intention prediction is crucial to traffic safety, which is a challenging task in real traffic scenarios. Traditional methods infer the intention of pedestrians to cross by predicting their future movements based on the observed trajectories in history. The performance of those methods is limited due to insufficient features and sources of information. To address those limitations, we propose a ViT-based model which incorporates multi-modal data to predict the pedestrian crossing intention. Specifically, the proposed model takes into consideration the visual information, poses, bounding box coordinates and action annotations, and gradually fuses those features for the final prediction. Besides, different data processing methods are designed based on the corresponding characteristics of different modalities to make full use of each type of data. Extensive ablation studies are conducted to show the performance of temporal modelling and feature fusion.

Humanlike Decision and Motion Planning for Expressway Lane Changing Based on Artificial Potential Field

The autonomous vehicles (AVs) need to share the driving environment with the human driving vehicles (HDVs) on expressway in the future. The non-humanlike lane changing (LC) behavior of AVs can mislead human drivers, which brings potential risks. Stronger humanlike ability requires a more complex algorithm. However, the requirement of the on-board vehicle computation resources limits the humanlike ability of LC algorithms. In this context, considering environmental risks, driver speed requirements and driver focus shifting process, this paper proposes a new type of LC algorithm based on artificial potential field (APF) aiming to improve the humanlike ability. The coupling relationship between the longitudinal and the lateral potential field forces is analyzed theoretically based on environmental risk APF. Based on the relationship, we proposed a conversion mechanism of the driver between the lateral and longitudinal potential field forces to mimic the driver’s speed requirement. Then a target lane selection strategy is designed to trade-off the driver’s speed requirement and safety between multiple lanes. Finally, to mimic the driver’s focus transfer process, the lateral forces of the ego vehicle are analyzed theoretically to further propose a moving virtual lane lines algorithm to build the lateral space varying potential field. The algorithm proposed in this paper is validated by comparing with other LC algorithms in the real traffic scenarios. The results indicate that the proposed algorithm has a better ability to mimic human driver’s LC decision-making and provides a smoother and safer humanlike trajectory.

Hybrid Genetic Firefly Algorithm-Based Routing Protocol for VANETs

Vehicular Adhoc Networks (VANETs) are used for efficient communication among the vehicles to vehicle (V2V) infrastructure. Currently, VANETs are facing node management, security, and routing problems in V2V communication. Intelligent transportation systems have raised the research opportunity in routing, security, and mobility management in VANETs. One of the major challenges in VANETs is the optimization of routing for desired traffic scenarios. Traditional protocols such as Adhoc On-demand Distance Vector (AODV), Optimized Link State Routing (OLSR), and Destination Sequence Distance Vector (DSDV) are perfect for generic mobile nodes but do not fit for VANET due to the high and dynamic nature of vehicle movement. Similarly, swarm intelligence routing algorithms such as Particle Swarm Optimization (PSO) and Ant Colony Optimization (ACO) routing techniques are partially successful for addressing optimized routing for sparse, dense, and realistic traffic network scenarios in VANET. Also, the majority of metaheuristics techniques suffer from premature convergence, being stuck in local optima, and poor convergence speed problems. Therefore, a Hybrid Genetic Firefly Algorithm-based Routing Protocol (HGFA) is proposed for faster communication in VANET. Features of the Genetic Algorithm (GA) are integrated with the Firefly algorithm and applied in VANET routing for both sparse and dense network scenarios. Extensive comparative analysis reveals that the proposed HGFA algorithm outperforms Firefly and PSO techniques with 0.77% and 0.55% of significance in dense network scenarios and 0.74% and 0.42% in sparse network scenarios, respectively.

Toward Driving Scene Understanding: A Paradigm and Benchmark Dataset for Ego-Centric Traffic Scene Graph Representation

Recently, beyond object detection and segmentation, high-level understanding of autonomous driving scenarios is attracting increasing attention. And traffic scene graph has become a potential way to model and represent the cognitive semantics of driving scenarios. However, it remains a challenging problem to construct traffic scene graph due to the complexity of driving scenes and the lack of benchmark datasets with various traffic entities and their relationships. In this paper, we propose a novel driving scene understanding paradigm, which can explicitly model traffic entities and the relationships between them from the view of ego-vehicle. Meanwhile, based on parallel vision and our paradigm, we propose an ego-centric Traffic Scene Graph dataset (TSG-451), which contains 451 images, 2,266 entities, and 4,272 relationships in real and artificial traffic scenarios. Furthermore, through qualitative and quantitative analysis, we validate the advancement of our proposed paradigm and dataset. Further investigation shows that our work can more comprehensively represent traffic scenarios and provide more fine-grained detailed semantics benefit to ego-vehicle’s decision-making.

Multi-Intersection Traffic Optimisation: A Benchmark Dataset and a Strong Baseline

The control of traffic signals is fundamental and critical to alleviate traffic congestion in urban areas. However, it is challenging since traffic dynamics are complicated in real-world scenarios. Because of the high complexity of the optimisation problem for modeling the traffic, experimental settings of existing works are often inconsistent. Moreover, it is not trivial to control multiple intersections properly in real complex traffic scenarios due to its vast state and action space. Failing to take intersection topology relations into account also results in inferior solutions. To address these issues, in this work we carefully design our settings and propose a new dataset including both synthetic and real traffic data in more complex scenarios. Additionally, we propose a novel baseline model with strong performance. It is based on deep reinforcement learning with an encoder-decoder structure: an edge-weighted graph convolutional encoder to excavate multi-intersection relations; and an unified structure decoder to jointly model multiple junctions in a comprehensive manner, which significantly reduces the number of the model parameters. By doing so, the proposed model is able to effectively deal with the multi-intersection traffic control optimisation problem. Models are trained/tested on both synthetic and real maps and traffic data with the Simulation of Urban Mobility (SUMO) simulator. Experimental results show that the proposed model surpasses multiple competitive methods. The traffic data and the code can be found at <uri xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">https://git.io/JPdU1</uri> .

Performance Evaluation of Camera-Based Time to Collision Calculation with Different Detectors&amp;Descriptors

Nowadays, the demand for producing and using autonomous vehicle is increasing. Due to the latest developments in technology, the capabilities of these vehicles in accident prevention are increasing. As a result of the accuracy of these capabilities, it is very important because it is human life. In today’s technology, the collision time calculation called TTC (Time to Collision) can be done in two different ways. The first of these methods is lidar-based calculation. In this paper TTC will be calculated using the camera-based method with different combinations of detectors and descriptors. Pros and cons of these methods will be discussed. The aim of this paper is to expose an exacting performance for related methods, especially its diverse combinations are used matching. In these experiments images are used for 10 images taken from real time traffic scenario of preceding vehicle. This paper includes seven methods for detectors and 6 methods for descriptors. These detectors and descriptors are used in 42 different combinations. The analysis includes four parameters such as total keypoint detection, total matches, total time in ms and performance ratio which is total matches divided by total time.

Proposal and Investigation of a Convolutional and LSTM Neural Network for the Cost-Aware Resource Prediction in Softwarized Networks

Resource prediction algorithms have been recently proposed in Network Function Virtualization architectures. A prediction-based resource allocation is characterized by higher operation costs due to: (i) Resource underestimate that leads to quality of service degradation; (ii) used cloud resource over allocation when a resource overestimate occurs. To reduce such a cost, we propose a cost-aware prediction algorithm able to minimize the sum of the two cost components. The proposed prediction solution is based on a convolutional and Long Short Term Memory neural network to handle the spatial and temporal correlations of the need processing capacities. We compare in a real network and traffic scenario the proposed technique to a traditional one in which the aim is to exactly predict the needed processing capacity. We show how the proposed solution allows for cost advantages in the order of 20%.

Hardware Limitations to Secure C-ITS: Experimental Evaluation and Solutions

Cooperative Intelligent Transportation Systems (C-ITS) improve driving experience and safety through secure Vehicular Ad-hoc NETworks (VANETs) that satisfy strict security and performance constraints. Relevant standards, such as the IEEE 1609.2, prescribe network-efficient cryptographic protocols to reduce communication latencies through a combination of the Elliptic Curve Qu-Vanstone (ECQV) implicit certificate scheme and the Elliptic Curve Digital Signature Algorithm (ECDSA). However, literature lacks open implementations and performance evaluations for vehicular systems. This paper assesses the applicability of IEEE 1609.2 and of ECQV and ECDSA schemes to C-ITSs. We release an open implementation of the standard ECQV scheme to benchmark its execution time on automotive-grade boards. Moreover, we evaluate its performance in real road and traffic scenarios and show that compliance with strict latency requirements defined for C-ITS requires computational resources that are not met by many automotive-grade embedded hardware platforms. As a final contribution, we propose and evaluate novel heuristics to reduce the number of signatures to be verified in real C-ITS scenarios.

A non-linear disturbance observer-based cooperative adaptive cruise control for real traffic scenarios

Cooperative adaptive cruise control (CACC) has a strong potential to improvise highway traffic capacity and ease traffic disturbances. Extensive exploration is not carried out in the area of CACC for a cut-in maneuver. Contemporary control strategies proposed for CACC cannot regulate the peaking of control input and thus the acceleration/deceleration of following vehicles when applied for various real traffic scenarios. This paper aims to develop a non-linear disturbance observer-based sliding mode control to control a CACC system for various traffic scenarios. The proposed observer estimates the uncertainty present in the actuator dynamics and the preceding vehicle’s acceleration as the lumped disturbance at the same time, it adjusts the observer gain to alleviate the peaking of control input. The stability of individual vehicles and the string stability of vehicle platoon are derived The performance of the proposed scheme is validated with various traffic scenarios, that is, cut-in maneuver, cut-out maneuver, and non-zero initial conditions. The effectiveness of the proposed scheme is demonstrated by comparing it with a linear disturbance observer-based control.