Autonomous Driving (AD) has garnered significant attention in recent years across multiple domains. Despite notable advancements in algorithms and hardware, large-scale deployment of autonomous vehicles remains constrained by the lack of adequate real-time guarantees. This article comprehensively investigates real-time assurance challenges in AD systems from theoretical and practical perspectives. First, we introduce foundational concepts of real-time systems and analyze common modeling approaches, including the multi-rate DAG and processing chain DAG models. We then delve into the task scheduling and communication mechanisms of three representative middleware systems–ROS2, Cyber, and ERDOS–and the seL4 operating system. Our analysis reveals their underlying design philosophies and optimization strategies for real-time performance. The findings highlight the critical difficulty of meeting stringent real-time requirements in autonomous systems. By offering insights into current limitations and opportunities for improvement, this article aims to establish a deeper understanding of real-time assurance issues and encourage greater focus on system-level guarantees within the AD community.

ABSTRACT This study applies text mining to South Korean news data to identify weak signals related to elderly drivers and analyze their policy implications. A total of 2264 news articles published between January 2024 and March 2025 were collected, and keyword-based weak signals were identified. Word network analysis interpreted relationships between weak signals and co-occurring terms. The results revealed five signal categories—acceleration/brake/activation, free, black box, sidewalk, and system—corresponding to driving errors, incentive-based policies, accident documentation, pedestrian safety, and administrative structures. The findings highlight the need for technical measures to prevent pedal misapplication, mobility-related support for those ceasing to drive, recording devices for accident analysis, protection infrastructure against sidewalk intrusion, and conditional licensing and integrated management systems for elderly drivers. Recognizing elderly driver-related accidents as a structural risk, this study provides insights for traffic safety and criminal justice policy using a weak signal detection approach grounded in public discourse.

The problem of the practical implementation of mathematical models of optimal motion of an electric vehicle with an AC traction electric drive in driver decision support systems is considered. The necessity of the equivalencing of complex dynamic models to ensure real-time calculations is substantiated. A method of context-dependent adaptive model reduction is proposed, which involves automatic switching between simplified equivalent models depending on the current driving mode of the vehicle: a horizontal section, ascent, descent or turn. Criteria for the equivalence of reduced models are formulated based on minimising the integral error of the velocity trajectory and energy consumption. A hybrid model structure with switching logic is developed, which ensures the continuity of the motion trajectory when changing modes. Computer modelling has been performed in the MATLAB/Simulink environment to compare the computational complexity of the full and equivalent models. The modelling results confirm a four- to sixfold reduction in computation time while maintaining the accuracy of motion parameter prediction at a level of 2% to 5% relative error. The conditions for the applicability of each of the equivalent models were determined depending on the road profile and dynamic movement characteristics. An algorithm for integrating equivalent models into a decision support system software application was proposed, with the possibility of further expanding functionality through intelligent prediction algorithms. The prospects for applying the developed models to the creation of adaptive electric vehicle control systems based on artificial intelligence methods are outlined. The results obtained can be used in the design of intelligent decision support systems for electric vehicle drivers, aimed at improving energy efficiency, safety and adaptability of movement in real operating conditions.

Autonomous driving systems (ADS) have been an active area of research, with the potential to deliver significant benefits to society. However, before large-scale deployment on public roads, extensive testing is necessary to validate their functionality and safety under diverse driving conditions. Therefore, different testing approaches are required, and achieving effective and efficient testing of ADS remains an open challenge. Recently, generative AI has emerged as a powerful tool across many domains and is increasingly applied to ADS testing due to its ability to interpret context, reason about complex tasks, and generate diverse outputs. To understand its role in ADS testing, we systematically analyzed 92 relevant studies and synthesized their findings into six major application categories, primarily centered on scenario-based testing. We further evaluated the effectiveness of these approaches and compiled a comprehensive set of resources, including 37 datasets, 17 simulators, 61 ADS systems, 141 metrics, and 100 benchmarks used for evaluation. In addition, we identified 27 limitations in existing research and outlined six major directions for future work. This survey provides an overview and practical insights into the use of generative AI for testing ADS, highlights existing challenges, and outlines future research opportunities in this rapidly evolving field.

Autonomous Driving Systems (ADS) are safety-critical. Abundant and various driving scenarios are required to train accurate and robust models, and comprehensively test each module of autonomous driving systems (i.e., perception, tracking, prediction, planning, and control modules). However, collecting driving scenario data from the real physical world is expensive and inefficient. Most existing works generate simulated driving scenarios by varying the behaviors of dynamic objects on simple road networks (e.g., highways), while the influence of roadside structures and scenarios with complex road networks are not considered. This paper proposes a novel driving scenario generation approach, Automated Scenario Crafting (AutoSceCraft), to automatically produce abundant driving scenarios containing various road networks, traffic rules, roadside structures, and dynamic objects at low cost. To validate the effectiveness and efficiency of our proposed framework, AutoSceCraft is integrated into three popular driving simulators, including SMARTS, esmini, and CARLA. Numerical experiments and scenario visualization results show that AutoSceCraft can generate effectively and efficiently various driving scenarios from scratch for testing and training various modules (including perception, prediction, and planning modules) within autonomous driving systems.

Simulation-based testing is essential for evaluating the safety of Autonomous Driving Systems (ADSs). Comprehensive evaluation requires testing across diverse scenarios that can trigger various types of violations under different conditions. While existing methods typically focus on individual diversity metrics, such as input scenarios, ADS-generated motion commands, and system violations, they often fail to capture the complex interrelationships among these elements. For instance, identical motion commands can produce different collision risks in varying scenes, and the same collision may result from different commands under different scenarios. This oversight leads to gaps in testing coverage, potentially missing critical issues in the ADS under evaluation. In this paper, we propose <monospace xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Causal-Fuzzer</monospace>, the first causality-aware fuzzing technique that enables efficient and comprehensive testing of ADSs by constructing causal graphs to model the interrelationships among scenarios, actions, and violations. Unlike existing methods that treat diversity metrics independently, we recognize these elements are causally interconnected and use their relationships to identify more diverse violations triggered by fundamentally different causal mechanisms. Specifically, <monospace xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Causal-Fuzzer</monospace> proposes (1) a causality-based feedback mechanism that quantifies the combined diversity of test scenarios by assessing whether they activate new causal relationships, and (2) a causality-driven mutation strategy that prioritizes mutations on input scenario elements with higher causal impact on ego action changes and violation occurrence to enable interpretable and efficient test generation. We evaluated <monospace xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Causal-Fuzzer</monospace> on an industry-grade ADS Apollo, with a high-fidelity simulator LGSVL. Our empirical results demonstrate that <monospace xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Causal-Fuzzer</monospace> significantly outperforms existing methods in (1) identifying a greater diversity of violations (96.5 violations on average, compared to 66.9 for the best baseline method), (2) providing enhanced testing sufficiency with improved coverage of causal relationships (13.6 unique sceneaction- violation patterns on average, compared to 8.6 for the best baseline method), and (3) achieving greater efficiency in detecting critical scenarios, strong robustness under noise conditions, and good generalizability across varying scenario complexities and violation types. Our source code and experimental results are available at <uri xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">https://sites.google.com/view/causal-fuzzer</uri>.

Automatic driving technology is a disruptive force to reshape the future travel and transportation system. Its development has undergone a fundamental transformation from a rule-based modular approach to a data-driven deep learning paradigm. This paper systematically summarizes the key role, frontier progress and core challenges of neural network as the core technology engine in the whole technology stack of automatic driving perception, decision-making, planning and control. Firstly, this paper summarizes the application foundation of convolutional neural network, cyclic neural network, transformer and graph neural network in the field of automatic driving; Then the representative work and implementation path in the core tasks of multi-sensor fusion 3D target detection, interactive behavior prediction, end-to-end driving are analyzed in detail. The analysis shows that although the neural network has greatly improved the performance ceiling of the system, its inherent black box characteristics, vulnerability to adversarial attacks, difficult to deal with the "long tail problem" and high computing costs are still the fundamental obstacles restricting its safe landing. Finally, this paper looks forward to the future research directions of multimodal large-scale model, causal reasoning, neural radiation field simulation and vehicle road coordination, aiming to provide a comprehensive and profound perspective of technology development for researchers in the field.

With self-driving technology, safety under intricate conditions for vehicles is indispensable. Image dehazing as a key element of visual sensing for self-driving vehicles is indispensable under poor environmental conditions. In complex traffic scenarios, visual perception systems must accurately recognize obstacles, traffic signs, and lane structures to ensure driving safety. However, under adverse weather conditions such as fog, haze, or sandstorms, image quality can severely degrade, leading to diminished detection performance and potential safety risks. The paper reviews the progression of image dehazing for self-driving vehicles with classical methods, deep learning-based methods, and unsupervised learning-based methods. The classical methods show efficiency but cannot cope with instability under intricate conditions. Deep learning-based methods, particularly CNNs and Transformers, have registered remarkable progress with issues of computational costs and data adaptability. Unsupervised learning-based methods reduce reliance on labeled data with issues of training instability and incomplete reconstruction. The paper also explores image dehazing as an integration with object detection and multi-sensor fusion, with future directions on lightweight design, data generalization, and multi-modal fusion.

Ride comfort is a critical issue in vehicle dynamics, as excessive vibrations adversely affect passenger comfort and human health. This paper presents a comparative performance analysis of a passive suspension system, fuzzy logic control (FLC), and a newly designed adaptive robust control (ARC) strategy applied to a nonlinear quarter-car suspension–seat–driver model. The primary objective is to improve ride comfort while maintaining vibration levels within accepted health criteria. First, the nonlinear dynamic model of the suspension–seat–driver system is established. The FLC structure and rule base are determined based on heuristic knowledge. Passive and FLC-based systems, while effective to some extent, suffer from limited adaptability to external disturbances and modeling uncertainties, slower convergence, and suboptimal vibration attenuation. The main contribution of this study is the design and implementation of a novel adaptive robust controller that effectively handles modeling uncertainties, external disturbances, and parameter variations. Different controller placement approaches within the system are also investigated. Numerical simulations are conducted under identical operating conditions for the uncontrolled system and all control strategies. The results demonstrate that although the FLC improves ride comfort compared to the passive system, the proposed ARC achieves the best overall performance, providing superior vibration attenuation, faster convergence, and enhanced robustness for nonlinear vehicle suspension systems. Quantitatively, the ARC reduces head acceleration RMS from 0.1693 m/s2 (passive) and 0.1422 m/s2 (FLC) to 0.0705 m/s2, and upper torso RMS from 0.1689 m/s2 (passive) and 0.1417 m/s2 (FLC) to 0.0703 m/s2, corresponding to approximately 58% reduction relative to passive and 50% improvement over FLC.

Ensuring the safety of Autonomous Driving Systems (ADS) at urban intersections remains challenging due to complex interactions between vehicles and traffic management infrastructure. This study validates an ADS equipped with connected perception using Infrastructure-to-Vehicle (I2V) communication within a combined virtual and hybrid testing approach. The validation follows the overall structure and methodology of the SUNRISE Safety Assurance Framework (SAF), which is applied in detail where required by the scope of the study. Five representative urban intersection scenarios, covering both nominal driving conditions and safety-critical edge cases, are evaluated using virtual simulations in MATLAB/Simulink (2014b) and hybrid experiments integrating OMNeT++ (5.7.1)/Veins (5.2)/SUMO (1.12.0) with real-world components. Key Performance Indicators (KPIs) related to safety, decision-making, longitudinal control, passenger comfort, and V2X communication performance are analyzed. The results show strong consistency between virtual and hybrid testing, with ego vehicle speed deviations below 2 km/h and trigger distance differences under 3 m. V2X communication achieves a near-perfect Cooperative Awareness Message (CAM) delivery ratio, with an average latency of approximately 142 ms. While this latency remains within the tolerance of the deployed ADS, the overall end-to-end delay highlights opportunities for further optimization. The study demonstrates how the SUNRISE SAF can effectively structure ADS validation, identifies critical scenarios such as right-of-way violations by non-priority obstacles, and provides insights into improving connectivity handling and low-speed braking behavior for Cooperative, Connected, and Automated Mobility (CCAM) systems in urban environments.

ABSTRACT In recent years, China’s autonomous driving technology has achieved rapid advancement, placing it among the global leaders. Advanced Driver Systems (ADS) demonstrate tremendous potential to effectively reduce traffic accidents and casualties, playing a significant role in safeguarding fundamental human rights such as the right to life and health. To promote the standardised application and widespread adoption of autonomous vehicles, China has successively formulated and revised relevant laws and regulations. However, the existing legal framework still has several deficiencies, such as unclear attribution of liability, inadequate data processing rules, and the absence of legislation on intelligent safety for commercial vehicles, failing to fully adapt to the rapid technological development. Looking ahead, priority should be given to establishing a human rights-oriented legal framework for autonomous driving, clarifying attribution of liability for traffic accidents, improving data processing rules, and promoting the incorporation of active safety technologies in national mandatory standards to further strengthen the protection of citizens’ fundamental rights.

A Study on the Effective Factor of the Driver-Inititated Disengagement of Autonomous Driving System Through Real-road Field Operational Tests

To investigate the knowledge and expectations of abdominopelvic MR elastography (MRE) among radiologists from multiple institutions, and to rate the image quality of MRE maps. Radiologists from Beijing and Tianjin were invited to participate in an online survey. Full MRE maps of the uterus, prostate, pancreas, liver, and kidney from 93 research papers published between 2017 and 2024 were displayed for image quality rating, blinded to the MRE systems. Before and after the image-review session, the participants' knowledge and expectations about abdominopelvic MRE were investigated. Eighty-one radiologists finalized the survey, all but one from tertiary hospitals. Their knowledge of MRE mainly came from the literature and conferences, and only 29.6% had hands-on experience. Additional setup and scanning time, as well as patient safety and comfort, were the major concerns that may hinder MRE application. Parametric maps from two MRE systems were reviewed, and the pressurized air (PA) system exhibited consistently higher subjective image quality for all five abdominopelvic organs compared to the acoustic driver (AD) system (all p < 0.05). The participants' knowledge about MRE improved after the image-review session, whereas their expectation declined regarding its utility in abdominopelvic tumor detection, characterization, staging, and treatment monitoring. Technical unfamiliarity with MRE was high, with ignorance rates ranging from 42.0% to 64.2%. There was limited knowledge and availability of MRE even in tertiary hospitals in China's capital region. Currently available MRE systems offer varying levels of anatomical details, and the PA system might be more advantageous for applications in abdominopelvic oncology. Question The radiologists' current knowledge and expectations of abdominopelvic MRE remained poorly investigated, and the impact of MRE image quality was unknown. Findings There was limited knowledge and availability of MRE in China's capital region, and MRE systems didn't consistently achieve sufficient image quality for abdominopelvic oncology. Clinical relevance Although MRE holds promise for abdominopelvic oncology, its availability and familiarity among radiologists remain low, and the image quality of MRE maps is sometimes insufficient. System optimization and targeted education are essential for future integration into clinical practice.

Ensuring safe operation of autonomous vehicle: A comprehensive survey of Operational Design Condition.

Vehicles with automated driving systems (ADS) and those with advanced driver assistance systems (Level 2 ADAS) operate in various locations in the US. The Federal Highway Administration (FHWA) has collected crash data for ADS and Level 2 ADAS–equipped vehicles. Understanding the safety benefits of these highly automated vehicles would provide the FHWA insights and guidance on improving their safety. Thus, this study applied Bayesian networks to evaluate the severity of ADS and Level 2 ADAS–involved crashes. Specifically, the study intends to determine the safety benefits of ADS over Level 2 ADAS in various traffic and roadway conditions. Two models focused on different categories of injury severity scale (KABCO) were developed. Results indicated that, overall, ADS-involved crashes are relatively less severe compared with Level 2 ADAS–involved crashes. On average, the probability of KA and KABC crashes is about 12% and 11% lower for ADS-equipped vehicles than Level 2 ADAS–equipped vehicles. The safety assessment across various roadway facilities, traffic conditions, and environmental factors showed relatively similar performance of ADS and Level 2 ADAS for KABC and KA crashes. Higher safety benefits of ADS-equipped vehicles were observed at low-speed limit roadways, clear weather, and intersections, among others. On the other hand, the safety performance of ADS-equipped vehicles deteriorates on the freeway, especially at 96 km/h (60 mi/h) and above. Overall, the findings suggest that most cases of ADS-equipped vehicle crashes are relatively less severe. These findings may be crucial to safety practitioners and planners focusing on deploying ADS-equipped vehicles on their roadway networks.

• We examined hazard perception in naturalistic driving videos using eye tracking. • Pupil dilation increased ∼ 1s before hazard detection and peaked ∼ 2s after. • Pupil dilation was larger for True-Positive than for False-Positive responses. • Fixation duration increased ∼ 2s before hazard detection, especially for True-Positives. • Attention remained predominantly focal, with no significant ambient-focal shifts. Hazard perception is the ability to anticipate and respond to potentially dangerous traffic situations, which is an important aspect of driving competence. This paper analyses pupil size fluctuations, changes in fixation duration, and the dynamics of ambient/focal attention in a laboratory hazard-perception task to measure underlying attentional and cognitive mechanisms that occur when drivers detect hazards. In the task, licensed drivers (n = 42) watched videos of natural driving scenarios recorded through a dashboard camera while their eye movements were recorded. They were asked to subjectively detect hazards (via a key press), which were later classified as either True- or False-Positive responses. We analyzed the time before and after the decision about the presence of a traffic hazard. As predicted, the pupil size increased over time. The pupil response was stronger for True than False hazard responses. The significant difference in pupil size between True and False responses appeared shortly before the decision and persisted for at least three seconds after it. There was no statistically significant differences in fixation duration over time, but True-Positive responses were related with higher fixation duration compared to False positive decisions. The increase in fixation duration was greater for True-Positive compared to False-Positive decisions. The analysis of ambient and focal attention dynamics revealed that participants maintained focal attention before and after hazard detection. The results show the potential for monitoring oculometrics in assistive driver systems for the detection of distraction and hazard perception in real time.

Autonomous driving systems (ADSs) must be sufficiently tested to ensure their safety. Though various ADS testing methods have shown promising results, they are limited to a fixed vehicle characteristics setting (VCS). The impact of variations in vehicle characteristics (e.g., mass, tire friction) on the safety of ADSs has not been sufficiently and systematically studied. Such variations are often due to wear and tear, production errors and so on, which may lead to unexpected driving behaviours of ADSs. To this end, in this article, we propose a method, named SafeVar , to systematically find minimum variations to the original vehicle characteristics setting, which affect the safety of the ADS deployed on the vehicle. To evaluate the effectiveness of SafeVar , we employed two ADSs and conducted experiments with two driving scenarios. Results show that SafeVar , equipped with NSGA-II, generates more critical settings that put the vehicle into unsafe situations, as compared with the baseline algorithm. We also identified critical vehicle characteristics and reported to which extent varying their settings put the ADS vehicle into unsafe situations.

Reading is a cognitively challenging skill that is used frequently in daily life, involving various visual factors. Previous studies have shown the impact of visual impairments on reading. The implementation of augmented reality (AR) is relatively new in ophthalmology. Studies have applied AR Head Mounted Displays (HMD) in vision enhancing systems for older drivers or as an instrument to compensate for metamorphopsia; however, the usability of an AR-HMD in ophthalmologic research has received little attention. The aim of this study was to assess the usability of AR-HMD for reading studies. Participants were randomly assigned to read aloud four of five texts with a similar level of reading skill as fast and accurately as possible, as displayed either on a PC monitor or on the Varjo XR4 AR HMD. Each test consisted of two reading blocks: a pilot for the participant to get familiar with the setup and the test with the recorded reading performance (4 blocks, 2 × 2). Reading speed (words/min) were recorded in both displays and eye movements were tracked for the reading experiment in the PC. At the end of the study, participants were asked about comfort, the adjustment of the AR-HMD and challenges experienced during the experiment. All participants (n = 31) successfully completed all blocks. There was no significant difference in the speed at which participants read the pilot and test reading texts in either device; however, there was a significant difference between the reading speeds assessed with the PC and with the AR HMD (p = 0.003). The average saccade amplitude and average peak velocity were strongly correlated. Individual mean saccade amplitudes ranged from 1.47° to 1.65° and the mean of the mean fixation durations were calculated at 276 - 288 ms. Common complaints with the AR HMD included its weight and tightness and some participants reported an unnatural visual experience during use. This study demonstrated the potential usability of an AR HMD device for reading experiments. Future uses of AR-HMD could leverage its advanced technical capabilities (eye and head tracking) for complex visual behavioral studies, realistic simulation of ophthalmic diseases, and objective assessment of driving capability in real-world scenarios.

Facial Trauma Prevention: Harnessing the Power of Autonomous Driving Systems.

Simulation models for perception sensors are integral components of automotive simulators used for the virtual validation of Autonomous Driving Systems (ADS). They also function as powerful tools for generating synthetic datasets to train deep learning-based perception models. LiDAR is a widely adopted perception sensor in ADS due to its high precision in 3D environment scanning. However, developing realistic LiDAR simulation models is a significant technical challenge. Unrealistic simulations can result in a large gap between the synthesised and real-world point clouds, limiting their effectiveness in perception and planning modules. Recently, deep generative models have emerged as promising solutions to synthesise realistic sensory data. However, for LiDAR simulation, deep generative models have been primarily hybridised with conventional algorithms, leaving unified generative approaches largely unexplored in the literature. Addressing this gap, we propose a unified generative framework to enhance LiDAR simulation fidelity. Our proposed framework projects LiDAR point clouds into depth-reflectance images via a lossless transformation, and employs our novel Controllable Lidar point cloud Generative model, CoLiGen, to translate the images. We extensively evaluate our CoLiGen model, comparing it with the state-of-the-art image-to-image translation models using various metrics to assess the realness, faithfulness, and performance of a downstream perception model. Our results show that CoLiGen exhibits superior performance across most metrics. The dataset and source code for this research are available at https://github.com/hamedhaghighi/CoLiGen.git.