Search

• We investigated a motion sickness mitigation method during a real car drive. • Vibrotactile cues announcing upcoming driving manoeuvres mitigate motion sickness. • Auditory cues are less effective than vibrotactile cues in mitigating sickness. • Passengers express greater preference for vibrotactile cues. Car passengers experience much more car sickness than car drivers. We assume that this is because drivers can better anticipate the car's motions. Does helping passengers to anticipate the car's motions then mitigate car sickness? Indeed, laboratory studies have shown that anticipatory cues which announce one-dimensional motions of a linear sled mitigate sickness to a small extent. Does this mitigation generalize to real car driving? We tested this in a car ride on a test track along a trajectory involving lane changes, accelerations, and decelerations. We show that vibrotactile cues mitigated car sickness in passengers. Auditory cues were less effective. The mitigating effect of the vibrotactile cue was considerable: a 40% decrease in car sickness symptoms, a larger effect than we found in the laboratory. Automated vehicles can predict their own motion very well. They could thus provide vibrotactile cues to mitigate car sickness in their passengers. [ABSTRACT FROM AUTHOR]

In the past few years, the lane detection technique has become a key factor for autonomous driving systems and self-driving cars on the road. Among the various vehicle subsystems, the lane detection module is one of the essential parts of the Advanced Driver Assistance System (ADAS). Conventional lane detection approaches use machine vision algorithms to find straight lines in road scene images. However, it is challenging to identify straight or curved lane markings in complex environments. To deal with this problem, this paper presents a lane detection technique based on deep learning. It is combined with a 3D convolutional network, so the temporal information is added to the network architecture. Using the front camera images, the system can immediately detect the lane marking information ahead. Moreover, we propose an approach for improvement by adding the time axis to the network architecture. In addition to using 3D-ResNet50, the temporal convolution and spatial convolution are separated for processing. The accuracy is 91.34% improved after adding time and space split convolution with LeakyReLU. The experiments carried out using real scene images have demonstrated the feasibility of the proposed technique for applications to various complex scenes. [ABSTRACT FROM AUTHOR]

No study has systematically investigated the public's perceptions of cyber risks and their relationship with the acceptance of fully Connected and Automated Vehicles (CAVs). To address this knowledge gap, we developed a conceptual model and investigated the impact of the cyber-emulated risks (cyberattack, safety risk, connectivity risk, privacy risk, and performance risk) that may influence the adoption of CAVs. We tested the proposed model using structural equation modelling with a nationally representative sample of 2062 adults from the US, UK, New Zealand, and Australia. The results indicate that perceived cyberattacks had a significant but marginally neutral effect on usage intent, illustrating the acceptance of technical risk with CAVs. This finding challenges the commonly held belief that cyberattacks negatively influence the adoption of products and technology in other product development fields, such as information technology. CAV cyberattacks elevate concerns about safety, connectivity, privacy, and performance risks. Interestingly, connectivity risk had no significant impact on CAV's behaviour intention, but mediation analysis showed it indirectly affects CAV's acceptance through privacy and performance risks. Regarding socio-demographic and technological attributes, participants of older age, middle income, low-middle education, high cybersecurity knowledge and AV understanding exhibit high anxiety about CAV cyberattacks. The results hold significant policy implications, suggesting the need for tailored strategies in enhancing the cybersecurity of CAVs to ensure their successful adoption and deployment. The findings of this study aim to enhance the quality of transport policy and bridge the gap between theory and practice in addressing cyber risks in the transport sector. • Developed a conceptual model with five theoretical constructs to assess the perceived cyber risks in AV adoption. • A nationally representative sample of 2062 adults from the US, UK, NZ, and AU was analysed to test the model. • Cyberattacks had a significant but marginally neutral effect on usage intent. • Older and middle-income participants exhibit high anxiety about CAV cyberattacks. • Participants with high cyber or AV knowledge were more worried about CAV cyberattacks. [ABSTRACT FROM AUTHOR]

• Stated preference survey among 1,000 respondents in the Philadelphia and the Seattle metro areas, two areas with a wide range of built environments and pedestrian behavior. • Descriptive data analysis shows pedestrians felt less safe with driverless cars on the road. • Pedestrians preferred disabling driverless function and capping speeds of driverless cars in urban core to increasing enforcement of jaywalking laws. • Respondents' current crossing behavior might serve as an indicator of their cross behavior in the presence of driverless cars. • Multilevel binomial logit shows pedestrians were more likely to jaywalk in the presence of driverless cars in high visibility and low traffic volume, and were deterred by bigger, faster streets. When interacting with driverless cars, pedestrians might be emboldened to jaywalk knowing the vehicles will slow down for them automatically. Increased jaywalking might disrupt traffic flow and pose challenges for traffic law enforcement. Few studies examined pedestrians' attitudes toward driverless cars' safety implication, management strategies for the interactions between pedestrians and driverless cars, and how pedestrians would expect their street crossing behavior to change in the presence of driverless cars. We conducted an online stated preference survey of 1,000 residents in the Philadelphia and the Seattle metropolitan areas to investigate respondents' attitudes toward driverless cars and their preferences of jaywalking under different driverless car saturation levels. Through descriptive data analysis and multilevel binomial logit analysis, we found that overall pedestrians felt less safe with driverless cars on the road. Pedestrians preferred limiting the speeds of driverless cars or disabling driverless functions in areas with heavy pedestrian traffic to increasing enforcement for jaywalking laws. Multilevel binomial logit analysis showed pedestrians were more likely to jaywalk in the presence of driverless cars in high visibility and low traffic volume, and were deterred by bigger, faster streets. Respondents' current crossing behavior might serve as an indicator of their crossing behavior in the presence of driverless cars. Our findings remind cities that sound street design strategies could help to ensure safer interactions between pedestrians and driverless cars and greater acceptance of driverless cars on urban streets. [ABSTRACT FROM AUTHOR]

In this research article, authors have engineered the model to control soft material robotic car with single and multi-hand gestures based on Artificial Intelligence and Machine Learning approach. A user centric representation of this features is obtained by using hand detection and hand gesture recognition mechanism. Authors have assigned the desire hand gestures to move the soft material robotic car in 360 degrees. In this research work, different gestures of human hands are simulated in terms of landmarks using Mediapipe. Each combination of hand landmark performs different actions to control the robotic car in multiple directions. In this research work, the soft material robotic car has been controlled using single hand as well as multi hands gesture-based approach. Our motive of this research work is to reduce human efforts by controlling the soft material robotic car through hand gestures based on AIML techniques. The proposed CNN model has yielded high accuracy as compared to baseline ML algorithms such as KNN, Decision Tree, SVM etc. It is also observed that CNN model remains ahead of other ML algorithms for both single hand and multi-hand gesture based robotic car control. The performance of proposed CNN model for controlling robotic car has been represented in term of ROC curve using validation and test data set. The average performances of single hand gesture and multi hand gestures using proposed CNN are found as 97% and 98% respectively. [ABSTRACT FROM AUTHOR]

Sentiment Analysis (SA) is a crucial task in understanding public opinions and perceptions towards emerging technologies. In this study, we focus on SA for a self-driving car dataset as it provides valuable insights into public perceptions and opinions towards a transformative technology. The dataset consists of textual reviews associated with sentiment labels, providing insights into how people perceive self-driving car technology. Our objective is to analyze the sentiments expressed in these reviews using Deep Learning (DL) models, namely, Artificial Neural Network (ANN), Convolutional Neural Network (CNN), Long Short-Term Memory (LSTM), Bidirectional LSTM (BiLSTM), Gated Recurrent Unit (GRU), and Bidirectional GRU (BiGRU). We compared our results with an existing technique in the field of self-driving car sentiment classification, that implemented various Machine Learning (ML) and DL models, including Support Vector Machines (SVM), Naïve Bayes (NB), Logistic Regression (LR), Random Forest (RF), CNN, and LSTM. In our study, we expanded upon this research by evaluating the performance of ANN, BiLSTM, GRU, and BiGRU. Results reveal that BiLSTM, GRU, and BiGRU exhibit superior performance in sentiment classification within the self-driving car dataset. These findings offer valuable insights into public sentiment towards self-driving cars, contributing significantly to the advancement of SA techniques in the domain of autonomous vehicles. Additionally, the results are statistically tested and are statistically significant. [ABSTRACT FROM AUTHOR]

Path-tracking and lane-keeping tasks are critical to guarantee safety and navigation performance considerations for deploying autonomous cars. This paper presents a novel control framework for the path-tracking control of high-speed autonomous cars with structured uncertainties. This study introduces a nonlinear adaptive control system based on a fractional-order terminal sliding mode system while incorporating a novel Gaussian Nonsingleton type-3 fuzzy system (FOTSM-NT3FS). Therefore, the proposed controller is independent of the information about the ego vehicle's dynamic information, and instead, the dynamics are approximated through a developed NT3FLS. The developed control system exhibits robustness to measurement errors and faulty sensors, because the inputs to the NT3FS are uncertain. In order to guarantee the boundedness of the adaptation parameters, the σ − mod approach is employed. The Lyapunov stability theorem and Barbalat's lemma are used to ensure the uniform ultimate boundedness of the closed loop system and the convergence of tracking errors to the origin in finite time. High-fidelity co-simulations with CarSim and MATLAB are performed to verify the effectiveness of the proposed control scheme and are also compared to other reported methods in the literature. Based on the obtained results, the schemed controller exhibits competitive effectiveness in path-tracking tasks and strong efficiency under various road conditions, parametric uncertainties, and unknown disturbances. • Robust path tracking for uncertain-input cars, ensuring reliability despite errors. • Adaptive control with fuzzy integration for high-speed autonomy, precise maneuvering. • Stable navigation using Gaussian Nonsingleton type-3 fuzzy system, smooth tracking. • Efficient path tracking under uncertainties, outperforming existing methods. [ABSTRACT FROM AUTHOR]

With rapid development of self-driving vehicles, recent work in adversarial machine learning started to study adversarial examples (AEs) for perception of autonomous driving (AD). However, generating practical AEs for the perception module remains a significant challenge. Traditional adversarial attacks tend to focus on a single computer vision task, making it difficult to compromise multiple perception tasks such as object detection and segmentation simultaneously. Additionally, the limited computational resources available on-board and the necessity for online operation pose further obstacles to deploying adversarial attacks on real autonomous driving platforms. To address the aforementioned issues, we propose the Time-aware Perception Attack (TPA), which is a real-time cross-task adversarial attack for perception of autonomous driving. In particular, we propose a novel backbone based adversarial attack method to modify input images to approach Lipschitz Constant Point (LCP), which results in erroneous inferences for all the sub-models in perception module. The novel part of this work is proposing an efficient yet effective LCP approaching algorithm. Comparing to conventional LCP based attacks, which consume significant amount of computation resources and can be only applied on small DNNs, TPA generates AEs on an intermediate layer of surrogate backbone, significantly enhancing the cross-task transferability and accelerates the attack process. Evaluation results on Berkeley Driving Dataset 100k (BDD100k) show that, comparing to the state-of-the-art baselines, the proposed TPA achieves higher attack effectiveness and faster processing speed and outperforms the baselines by a large margin. • The proposed method generates AEs with high transferability across perception tasks. • Approaching LCP compromises layer vulnerability, affecting task-head performance. • Attacking middle layers improves the efficiency of generating adversarial examples. [ABSTRACT FROM AUTHOR]

• The deviations between measured information and true information are quantified by information uncertainty level. • A generic model is developed to describe the scenario where multiple information uncertainties exist. • The impacts of multiple information uncertainties on stability of connected automated vehicles flow are explored. • A novel car-following controller is developed to address the adverse effects of multiple information uncertainties. Connected automated vehicles have the capability to operate autonomously by monitoring real-time traffic information through on-board sensors, such as velocity and distance. However, no measurement can be perfect, and sensors are no exception, especially in challenging road and weather conditions, leading to the deviations between multiple information measured by vehicles and true information. Since the sizes of sensor detection errors are uncertain, we call this issue as multiple information uncertainties. This issue affects not only the normal operation of host vehicle directly but also the connected automated vehicular flow indirectly through wireless communication, resulting in the instability of car-following behavior and further deteriorating traffic congestion. So far, it is hard for us to obtain the repeatable, transferable, and even comparable results due to the lack of generic model. Therefore, this study develops a generalized model by using the uncertainty levels of multiple information to describe the dynamics of connected automated vehicles under the influence of sensor detection errors based on car-following theory. The theoretical and simulation-based investigations present a complete method to analyze the stability of traffic flow under multiple information uncertainties. Analytical results show that traffic stability will be reduced when the velocity measured by sensors is smaller than true velocity (i.e., negative uncertainty level of velocity information) or the headway monitored by sensors is bigger than real headway (i.e., positive uncertainty level of headway information), whereas the velocity and headway of equilibrium state will be enlarged. Otherwise, the opposite. These findings indicate that the impacts of multiple information uncertainties are double-edged swords, depending on the uncertainty levels of different information. To improve the adverse impacts of multiple information uncertainties on traffic stability, this study proposes a novel car-following controller and verifies its effectiveness. Overall, the present study provides a set of theoretical frameworks to investigate and improve traffic stability under multiple information uncertainties. All results contribute to enhancing the stability of traffic flow and further easing traffic congestion. [ABSTRACT FROM AUTHOR]

The technological advancements of Connected and Automated Vehicles (CAVs) are outpacing the current regulatory regime, potentially resulting in a disconnect between legislators, technology, and CAV stakeholders. Although many studies explore the regulatory requirements of operations of CAVs, studies on regulatory challenges specific to the cybersecurity of CAVs are also emerging and receiving lots of attention among researchers and practitioners. However, studies providing an up-to-date synthesis and analysis on CAVs regulatory requirements specific to cyber-risk reduction or mitigation are almost non-existent in the literature. This study aims to overcome this limitation by presenting a comprehensive overview of the role of key Intelligent Transportation Systems (ITS) stakeholders in CAV's cybersecurity. These stakeholders include road operators, service providers, automakers, consumers, repairers, and the general public. The outcome of this review is an in-depth synthesis of CAV-based ITS stakeholders by visualising their scope in developing a Cybersecurity Regulatory Framework (CRF). The study demonstrated the compliance requirements for ITS communication service providers, regulatory standards for CAVs automakers, policy readiness for CAVs customers and the general public who interact with CAVs, and the role of the CAVs Network Operator Centre in regulating CAVs data flow. Moreover, the study illuminates several critical pathways necessary in future for synthesizing and forecasting the legal landscape of CAV-based transportation systems to integrate the regulatory framework for CAV stakeholders. The paper's findings and conclusions would assist policymakers in developing a comprehensive CRF. CAV Cybersecurity Regulation, Stakeholder Roles, and Conceptualization of CAVs (traffic laws) breaches. [Display omitted] • The study offers a comprehensive overview of CAV Cybersecurity's Regulatory Framework. • Considers road operators, service providers, automakers, consumers, repairers and the public. • Emphasizes automaker regulatory requisites for cyber-safe CAV operation. • Proposes CAVs Network Operation Centre to regulate data flow. • Advocates eSafety Traffic Unit for regulating CAV operations. [ABSTRACT FROM AUTHOR]

Car manufacturers around the globe are in a race to design and build driverless cars. The concept of driverless is also being applied to any moving vehicle such as wheelchairs, golf cars, tourism carts in recreational parks, etc. To achieve this ambition, vehicles must be able to drive safely on streets stay within required lanes, sense moving objects, sense obstacles, and be able to read traffic signs that are permanent and even temporary signs. It will be a completely integrated system of the Internet of Things (IoT), Global Positioning System (GPS), Machine Learning (ML)/Deep Learning (DL), and Smart Technologies. A lot of work has been done on traffic sign recognition in the English language, but little has been done for Arabic traffic sign recognition. The concepts used for traffic sign recognition can also be applied to indoor signage, smart cities, supermarket labels, and others. In this paper, we propose two optimized Residual Network (ResNet) models (ResNet V1 and ResNet V2) for automatic traffic sign recognition using the Arabic Traffic Signs (ArTS) dataset. Additionally, the authors developed a new dataset specifically for Arabic Traffic Sign recognition consisting of 2,718 images taken from random places in the Eastern province of Saudi Arabia. The optimized proposed ResNet V1 model achieved the highest training and validation accuracies of 99.18% and 96.14%, respectively. It should be noted here that the authors accounted for both overfitting and underfitting in the proposed models. It is also important to note that the results achieved using the proposed models outperform similar methods proposed in the extant literature for the same dataset or similar-size dataset. [ABSTRACT FROM AUTHOR]

Cooperative, connected and automated mobility (CCAM) has a great potential to change drastically the mobility landscape and improve safety on the roads by assisting drivers to take the best decisions in given circumstances or by supporting autonomous driving features aboard the car, among other expected benefits. Yet, providing CCAM services poses not only technical but also business challenges, especially in cross-border environments. In this paper, we present a general techno-economic methodology intended first, to identify the cross-border ecosystem and the potential business models to enable CCAM services provision and second, to derive the Total Cost of Ownership (TCO) of deploying the required Vehicle to Infrastructure (V2I) and Vehicle to Network (V2N) infrastructure. As an illustration, we then apply this methodology to a canonical Cooperative Lane Merging (CLM) scenario in a representative border environment (i.e., the Brenner pass between Italy and Austria). On this occasion, the impact of Multi-access Edge Computing (MEC) placement on the deployment cost is also discussed and finally, a cost allocation model is developed to link these deployment costs to the key stakeholders within the identified business models. Results aim to give insights into the different deployment strategies in typical cross-border environments and settings. • This paper presents a comprehensive techno-economic analysis of providing CCAM services in cross-border environment considering both technical and business challenges. • Proposed business models based on the adoption of new 5G technologies namely network slicing and MEC to ensure seamless service provision. • Collaborative business model is the most promising but needs review of regulations. • The collaborative business model is particularly attractive when considering the implementation of passive and active network sharing strategies. • Cost modelling results provide TCO insights for 5G network deployment scenarios. [ABSTRACT FROM AUTHOR]

• The greater the car's lateral acceleration, the more severe the car sickness (CSR). • Unpredictability of the vehicle's path worsens symptoms. • Higher CSR are related to greater changes in physiological parameters. • Physiological changes reflect activation of the sympathetic nervous system. • Cardiac and respiratory measures explain 41 % of the variance in peak CSR. With the development of autonomous vehicles, car sickness may affect increasing numbers of car occupants. Car manufacturers have a real need to understand the causes of these symptoms, which occur mainly when car occupants are not engaged in a driving task. This study is the first to evaluate, in real driving conditions, the impact of lateral acceleration level and vehicle path predictability on car sickness incidence and severity, and the potential relationship with physiological changes. 24 healthy volunteers participated as front seat passengers in a slalom session inducing lateral movements at very low frequency (0.2 Hz). They were continuously monitored via physiological recordings and provided subjective car sickness ratings (CSR) after each slalom, using a 5-point likert scale. CSR reveal that (i) the greater the lateral acceleration and (ii) the less predictable the vehicle path, the more severe the car sickness symptoms in real driving conditions. An increase in several physiological parameters is also found simultaneously with higher CSR, demonstrating activation of the sympathetic nervous system. Moreover, the linear regression applied to our data suggests that these physiological parameters can be used to indicate car sickness severity. Moreover, the linear regression applied to our data suggests that the evolution of these physiological parameters may reflect the CSR level indicated by participants. [ABSTRACT FROM AUTHOR]

• Online tracking and real-time multi-object tracking and segmentation(MOTS) tracker. • Mask-guided re-identification(ReID) using bounding box information and mask together. • End-to-end learning combining ReID branch with instance segmentation network. • Evaluation of an experiment in comparison with state-ofthe-art methods. • Ablation study to validate mask-guided ReID branch and feature dimension. Applications of multi-object tracking and segmentation (MOTS) include autonomous driving and video surveillance systems. In these applications, the tracking system must be able to track in real-time and online. Recent studies to improve the tracking performance of MOTS have been actively conducted; however, most studies fail to consider online and real-time performance. This paper proposes a MOTS system that operates in real time using a deep learning–based model with a light backbone, and can be tracked online. Additionally, state-of-the-art trackers use an object's bounding box to extract re-identification (ReID) features, but include unnecessary background. Additionally, this causes confusion in accurately expressing the appearance features of an object, resulting in difficulty in matching. To solve this difficulty, instead of providing the features of the bounding box as the input of the ReID branch, we focus on expressing the object by providing the mask features of the object as an input. As a result of the mask-based ReID experiment, the association accuracy performance was higher than that of the existing bounding box–based ReID model, and among the KITTI MOTS benchmarks, it ranked second among models that can operate online. Our experiments show that background information causes ambiguous ReID matching in MOTS systems, and that object mask information is important for avoiding ambiguous matching. [ABSTRACT FROM AUTHOR]

The development of autonomous vehicles has brought a great impact and changes to the transportation industry, offering numerous benefits in terms of safety and efficiency. How ever, one of the key challenges that autonomous driving faces is how to make ethical decisions in complex situations. To address this issue, in this article, a novel trajectory prediction method is proposed to achieve ethical decision-making for autonomous driving. Ethical considerations are integrated into the decision-making process of autonomous vehicles by quantifying the utility principle and incorporating them into mathematical formulas. Furthermore, trajectory prediction is optimized using LSTM network with an attention module, resulting in improved accuracy and reliability in trajectory planning and selection. Through extensive simulation experiments, we demonstrate the effectiveness of the proposed method in making ethical decisions and selecting optimal trajectories. [ABSTRACT FROM AUTHOR]

Road traffic has become prominent in everyday living, impacting or disrupting services to people and daily routines. With the rise in automobile manufacturing and the frequency of vehicle crashes, human catastrophes, like fatalities, accidents, impairments, and destruction of property, are surging yearly. Vehicle collision detection has recently gained prominence in decreasing manually operated and autonomous vehicle fatalities. The concept of independent and self-driving cars relies on accurate object recognition, including pedestrians, vehicles, buildings, and other moving objects. Various object-detecting approaches have been proposed to help autonomous vehicles (AVs) achieve consistent, safe driving. Object prediction and detection have noticed numerous algorithmic changes that have improved speed and accuracy. In this study, I used a traffic dataset produced by a CARLA simulator to anticipate collisions using the Yolov7 model. I generated the dataset from a CARLA simulation bench in video sequences, manually annotated the frames, and used the deep learning algorithm Yolov7 to train them. The model predicts the collision a few seconds before it occurs in real-time. I implemented this framework to increase the safety of driving in self-driving vehicles. [ABSTRACT FROM AUTHOR]

• Conditional process modelling was used to get a more detailed insight into the connections among attitudes toward autonomous vehicles, acceptance of technology, previous experience, and gender. • The findings suggest that attitudes rather than gender or previous experience explain acceptance of self-driving cars. • Gender had a slight moderation effect on the relationship between attitudes and acceptance. This article investigates the acceptance of autonomous cars based on the role of attitudes toward autonomous vehicles, acceptance of technology, previous experiences, and gender. Using an online survey, which includes the Acceptance of Autonomous Vehicles (AVAS) and Technology Adoption Propensity scale (TAP), a sample of 1273 members of a university community was collected. Acceptance of using autonomous cars in a test drive and ordinary, real traffic scenarios, as well as the intention to buy one were measured via self-administered items. We used conditional process modeling to get a more detailed insight into the connections among these factors. The findings revealed that all four factors of attitudes towards autonomous vehicles (benefits in usefulness, benefits in situations, commonalities concerns, system concerns) but only the optimism factor of technology adaption propensity affected the acceptance. Dependency seemed to affect benefits in usefulness and the two concern variables. Gender differences are almost entirely explained away by the effects of attitudes. Previous experience had no significant effect in the model. [ABSTRACT FROM AUTHOR]

In this paper we describe the design and the ideas motivating a new Continual Learning benchmark for Autonomous Driving (CLAD), that focuses on the problems of object classification and object detection. The benchmark utilises SODA10M, a recently released large-scale dataset that concerns autonomous driving related problems. First, we review and discuss existing continual learning benchmarks, how they are related, and show that most are extreme cases of continual learning. To this end, we survey the benchmarks used in continual learning papers at three highly ranked computer vision conferences. Next, we introduce CLAD-C, an online classification benchmark realised through a chronological data stream that poses both class and domain incremental challenges; and CLAD-D, a domain incremental continual object detection benchmark. We examine the inherent difficulties and challenges posed by the benchmark, through a survey of the techniques and methods used by the top-3 participants in a CLAD-challenge workshop at ICCV 2021. We conclude with possible pathways to improve the current continual learning state of the art, and which directions we deem promising for future research. • A review of current continual learning benchmarks and their relations to one-another. • Proposal of two new real-world continual learning benchmarks. • Review of the most successful ICCV'21 challenge entries on the new benchmarks. [ABSTRACT FROM AUTHOR]

• We discuss the benefits of accurate detection of small objects like traffic signs and traffic lights in autonomous driving. • We analyze the practical limitations of the original YOLOv5 structure. • We propose novel architectural refinements to the same for improving its performance in the detection of small objects. • We perform extensive experimentation over the BDD100K, TT100K, and DTLD datasets. • We further evaluate the generalization ability of the proposed iS-YOLOv5 model in different road weather conditions. With the rapid advancements in the field of autonomous driving, the need for faster and more accurate object detection frameworks has become a necessity. Many recent deep learning-based object detectors have shown compelling performance for the detection of large objects in a variety of real-time driving applications. However, the detection of small objects such as traffic signs and traffic lights is a challenging task owing to the complex nature of such objects. Additionally, the complexity present in a few images due to the existence of foreground/background imbalance and perspective distortion caused by adverse weather and low-lighting conditions further makes it difficult to detect small objects accurately. In this letter, we investigate how an existing object detector can be adjusted to address specific tasks and how these modifications can impact the detection of small objects. To achieve this, we explore and introduce architectural changes to the popular YOLOv5 model to improve its performance in the detection of small objects without sacrificing the detection accuracy of large objects, particularly in autonomous driving. We will show that our modifications barely increase the computational complexity but significantly improve the detection accuracy and speed. Compared to the conventional YOLOv5, the proposed iS-YOLOv5 model increases the mean Average Precision (mAP) by 3.35% on the BDD100K dataset. Nevertheless, our proposed model improves the detection speed by 2.57 frames per second (FPS) compared to the YOLOv5 model. [ABSTRACT FROM AUTHOR]

In this paper, event-triggered model predictive control (EMPC) with adaptive artificial potential field (APF) is designed to realize obstacle avoidance and trajectory tracking for autonomous electric vehicles. An adaptive APF cost function is added to achieve obstacle avoidance and guarantee stability. The optimization problem for MPC is feasible by considering a special obstacle avoidance constraint. An event-triggered mechanism is proposed to reduce computational burden and ensure effectiveness of obstacle avoidance. Input and state constraints of autonomous electric vehicles are considered in both feasibility and stability by a robust terminal set. Effectiveness of both obstacle avoidance and trajectory tracking is shown by experimental results on autonomous electric vehicles. • A terminal set is designed to ensure stability under constraints and disturbance. • An adaptive force coefficient is designed to achieve stable tracking and avoiding. • An event-triggered mechanism considering obstacles is used to reduce computation. • A special constraint is developed to ensure the optimization feasibility. • Experimental results are given to show effectiveness of the EMPC strategy. [ABSTRACT FROM AUTHOR]

In the digital era, human and thing behavioral patterns have been merged, which leads to the need for trust management to secure the relationship among people and things (e.g., driverless cars). Due to the dynamism and complexity of digital environments, trust management depends largely on indirect trust to support its reasoning by building the reputation of trustees based on recommendations reflected in the feedback of sentiment and non-sentiment objects. However, different biases are still affecting the accuracy of indirect trust that reflects a collective trustworthiness belief or societal stereotypes. This work focuses on enabling indirect trust management by leveraging deep learning in combination with synthetic data for bias management. Specifically, this paper proposes a reputation model to support decision-making in trust management by minimizing bias in indirect trust information and fostering fairly the relationship among sentiment and non-sentiment objects. Our experimental results show that the synthetic data can significantly improve the classification accuracy in trust management. [ABSTRACT FROM AUTHOR]

Self-driving and connected cars are designed to make traffic safer while lowering risks and accidents. However, Security concerns about these vehicles limit their use and what they can accomplish. Cyber-attacks on these vehicles may have serious repercussions, including the loss of personal information as well as physical harm or even death. In this paper, we investigate intrusion detection on connected and autonomous vehicles using supervised machine learning techniques. The overarching goal is to develop a taxonomy for linked intrusion detection systems and supervised learning algorithms. To that end, we provide a thorough explanation of the concepts of intrusion detection systems and cyber-security attacks. Then, we demonstrate how machine learning can improve the security of future connected and autonomous vehicles by examining and evaluating how to protect sent and transferred data. Finally, a taxonomy based on these related works is offered. Based on a review of four well-known data sets in this field, we can conclude from this taxonomy that the classification performance of supervised learning algorithms is strong and encouraging. [ABSTRACT FROM AUTHOR]

The fourth industrial revolution (4IR or Industry 4.0), fueled by cutting-edge technologies such as autonomous cars, advanced robotics, business analytics, and artificial intelligence, is expected to transform all aspects of today's workforce. Yet very few empirical studies have been conducted about the future skills needed to cope with this new industrial wave. This study is an initial attempt to fill this knowledge gap through identifying relevant literature. It aims at presenting a holistic view of the extent of existing literature on 4IR skills using a bibliometric analysis. Four hundred and ninety-one documents were extracted from the Web of Science (WoS) database and analyzed using two bibliometric tools, namely Bibliometrix and VOSViewer. The skills required for the 4IR began earnestly at the WEF in 2016. Currently, Italy emerges as the leader in 4IR skills research even though its adoption into their national plan occurred in September 2016. Thematic research gaps in skills and future clusters remain relevant even in the current era. [ABSTRACT FROM AUTHOR]

• Compared with people who drive cars, motorcyclists have lower trust for human drivers. • Motorcyclists compared with car drivers are less negative towards self-drive cars. • Motorcyclists compared with car drivers have a higher sense of trust in self-drive cars. • Car drivers express more concerns and a wider range of concerns in self-driver cars. • Motorcyclists are more likely than car drivers to consider their own safety when balancing the safety of self-drive cars. • Motorcyclists believe that self-drive cars will recognise their personal safety, compared to human drivers. Self-drive vehicles have the potential to revolutionise transport systems, however their adoption by mainstream users is highly dependent on a range of factors including trust. Moreover, perceptions of self-drive cars are likely to be highly dependent on the lived-experience of different road user types. Motorcyclists have high impressions of road-vulnerability and distrust for other road users. This distrust stems from the evidence that the vast number of motorcycle crashes occur as a consequence of other-driver error. We therefore hypothesised that compared with car drivers, vulnerable road users such as motorcyclists would report high levels of other-driver distrust, but high levels of trust in self-drive cars. We conducted semi-structured interviews with motorcyclists and car drivers and found that motorcyclists are indeed distrustful of other drivers but have higher levels of trust for self-drive cars compared with car drivers, while car drivers have low levels of trust for self-drive cars, but high levels of trust for other drivers. Motorcyclists were also more likely to report a perspective of trust that centred around their own personal safety on the road. This study provides valuable insights into different road user's perceptions of AVs. [ABSTRACT FROM AUTHOR]

In this paper, a low cost method for input size reduction without sacrificing accuracy is proposed, which reduces required computation resources for both training and inference of deep convolutional neural network (DCNN) in the steering control of self-driving cars. Efficient processing of DCNNs is becoming prominent challenge due to its huge computation cost, number of parameters and also inadequate computation resources on power efficient hardware devices and the proposed method alleviates the problem comparing the state of the art. The proposed method introduces feature density metric (FDM) as criterion to mask and filter out regions of input image that do not contain adequate amount of features. This filtering method prevents DCNN from useless calculations belongs to feature-free regions. Compared to PilotNet, the proposed method accelerates overall training and inference phases of end-to-end (ETE) deep steering control of self-driving cars up to 1.3× and 2.0× respectively. [ABSTRACT FROM AUTHOR]

Correct environmental perception of objects on the road is vital for the safety of autonomous driving. Making appropriate decisions by the autonomous driving algorithm could be hindered by data perturbations and more recently, by adversarial attacks. We propose an adversarial test input generation approach based on uncertainty to make the machine learning (ML) model more robust against data perturbations and adversarial attacks. Adversarial attacks and uncertain inputs can affect the ML model's performance, which can have severe consequences such as the misclassification of objects on the road by autonomous vehicles, leading to incorrect decision-making. We show that we can obtain more robust ML models for autonomous driving by making a dataset that includes highly-uncertain adversarial test inputs during the re-training phase. We demonstrate an improvement in the accuracy of the robust model by more than 12%, with a notable drop in the uncertainty of the decisions returned by the model. We believe our approach will assist in further developing risk-aware autonomous systems. • A new method for uncertainty-based test set generation for autonomous systems. • Support training more robust DL models for autonomous driving. • Improved model performance in the existence of data perturbations and adversarial attacks. [ABSTRACT FROM AUTHOR]

Automated driving systems (ADSs) conceive an efficient and safe way of driving. The safety of ADSs depends on a precise object detector that needs to be upgraded continuously facing various environments. Massive annotations are required to utilize collected images of surroundings through vehicles and accommodate new environments. Auto labelling is one approach to alleviate such dilemma. To this end, we propose a novel Weakly Supervised Object Localization (WSOL) method which can localize objects precisely without detection annotations. This paper proposed Soft Guidance Module (SGM), Channel Erasing Module (CEM) and incorporate them into a multi-flow framework allowing the two mutually beneficial. Finally, experiments and visualizations are performed to evaluate our method on Stanford Cars, ILSVRC 2016 and CUB-200-2011 datasets. • WSOL is used to generate object annotation for upgrading automated driving systems. • Soft Guidance Module detects more complete object including the edge adaptively. • Channel Erasing Module promotes networks to focus on more object-related channels. • Multi-flow structure combines two erasing modules without additional parameters. [ABSTRACT FROM AUTHOR]

Self-driving cars not only solve the problem of navigating safely from location A to location B; they also have to deal with an abundance of (sometimes unpredictable) factors, such as traffic rules, weather conditions, and interactions with humans. Over the last decades, different approaches have been proposed to design intelligent driving systems for self-driving cars that can deal with an uncontrolled environment. Some of them are derived from computationalist paradigms, formulating mathematical models that define the driving agent, while other approaches take inspiration from biological cognition. However, despite the extensive work in the field of self-driving cars, many open questions remain. Here, we discuss the different approaches for implementing driving systems for self-driving cars, as well as the computational paradigms from which they originate. In doing so, we highlight two key messages: First, further progress in the field might depend on adapting new paradigms as opposed to pushing technical innovations in those currently used. Specifically, we discuss how paradigms from cognitive systems research can be a source of inspiration for further development in modelling driving systems, highlighting emergent approaches as a possible starting point. Second, self-driving cars can themselves be considered cognitive systems in a meaningful sense, and are therefore a relevant, yet underutilized resource in the study of cognitive mechanisms. Overall, we argue for a stronger synergy between the fields of cognitive systems and self-driving vehicles. [ABSTRACT FROM AUTHOR]

This study presented a novel liquid-cooled heat sink based on constructal theory. An experiment was conducted to investigate the influence of boundary conditions, such as the mass flow rate (m ̇) , on heat transfer rate (Q in) and pressure drop. Five cylinder heater cartridges were used in the experiment for 11 different mass flow rates (0.008292 < m ̇ (kg/s) < 0.03307). Through numerical simulation, the effects of changing the number of clusters on heat transfer and pressure drop were studied. The results showed that the optimal combination of pressure drop and Nusselt number occurs in four clusters. According to the results, increasing the number of clusters can increase the Nusselt number by up to 11.98% and 13.62% for the highest (m ̇ = 0.03307 kg/s) and lowest (m ̇ = 0.008292 kg/s) mass flow rates, respectively. This work may lay the foundation for creating the next generation of thermal management systems for compact heat sources, such as the CPU in a self-driving car, robots and high-performance computers (HPC). [ABSTRACT FROM AUTHOR]

Efficient and accurate 3D multiobject tracking (MOT) is essential for autonomous driving. However, existing methods still struggle with multiobject occlusion, in which interruptions or even loss of crucial tracked objects is likely. Therefore, a more intelligent object-aware anti-occlusion 3D MOT for autonomous driving, named OATracker, is developed in this paper. First, an updated 3D cascaded detection (3DCD) is adopted as the detection backbone, with incorporation of an updatable component based on the OpenPCDet framework to increase the longevity. Second, a reliable object state-constrained affinity (OSA) metric is designed that adaptively combines 3D geometric and motion metrics to dynamically measure objects' size and distance in both spatial and temporal dimensions. Third, an object-aware trajectory management (OTM) module is developed for trajectory management and status updating. Specifically, the object identity-aware adaptive tracking life manager (ATLM) allows the tracker to adaptively manage the trajectories of captured objects through a confidence metric for each objects, and the ghost-aware trajectory postoptimizer (GTP) associates the historical and detected motion states, further overcoming id-switch and fragmentation problems. Finally, experiments on the KITTI-Car tracking benchmark and practical autonomous driving car application platform show that the OATracker can achieve competitive and robust tracking performance even in occlusion conditions. [ABSTRACT FROM AUTHOR]

Accurate and reliable perception is essential for the safe operation of autonomous cars. The task of Light Detection And Ranging (LiDAR) panoptic segmentation plays a crucial role in predicting point-level classes and instance IDs within a 3D environment, facilitating a comprehensive understanding of the vehicle's surroundings. Existing approaches relying on a single LiDAR sensor often face limitations arising from challenges such as point sparsity and the absence of texture information. To overcome these challenges, the fusion of other sensors, such as cameras, can be employed. However, the development of models for panoptic segmentation utilizing sensor fusion methods remains an unexplored area. In this paper, we present the first LiDAR-camera fusion network for panoptic segmentation named "Panoptic-FusionNet." Our proposed network enhances feature maps through a feature fusion module that ensures precise geometric alignment of features extracted from both of sensors. We create tables that have accurately aligned point-voxel-pixel correspondence at multiple scales, and fuse feature maps by querying the table. The fused features are subsequently employed in a bottom-up network to predict semantic labels, instance centers, and offsets. Post-processing is then implemented in a class-adaptive manner, enabling class-wise clustering based on the general sizes of each class, thereby refining the final panoptic results. We evaluate the performance of the proposed network on both the SemanticKITTI and nuScenes datasets, which are large-scale datasets that provide synchronized LiDAR-camera data. In comparison to LiDAR-only baselines, Panoptic-FusionNet attains higher panoptic quality due to its geometrically accurate sensor fusion method, class-adaptive processing, and utilization of instance-augmented pre-trained weights. [ABSTRACT FROM AUTHOR]

Growing global research utilizes user acceptance models to investigate the public acceptance of automated vehicles (AVs). A growing body of literature suggests it is essential to recognize cultural differences that may influence people's decisions and the intention to use (AVs). While the influence of perceived safety on AVs adoption has been examined globally, it has often been overlooked in Australia. To address this knowledge gap, this study extended the Unified Theory of Acceptance and Use of Technology (UTAUT) model by incorporating perceived safety and socio-demographic factors in assessing behavioral intention for fully AVs in Australia. This study is the first in Australia to include perceived safety in the UTAUT model and look at how different factors like age, gender, experience, income, education, and travel habits affect people's intention to use technology. The model was evaluated with Structural Equation Modelling using a dataset of 804 respondents from Australia. Perceived Safety (PS) holds comparable importance to Social Influence (SI) and Facilitating Conditions (FC). Our analysis revealed that younger age groups exhibit a more substantial positive correlation between Performance Expectancy (PE) and Behavioral Intention (BI) compared to older age groups. Notably, there are significant distinctions in the impact of PS on BI between older and younger age groups, as well as between those with and without prior experience with AVs. Moreover, gender has a moderating effect akin to age in the PE-BI relationship. Our findings also reveal that age moderates the relationship between PE and BI, with younger individuals exhibiting less susceptibility to social influence compared to older counterparts. Gender also emerges as a moderator, affecting the relationship between FC and BI. Additionally, income moderates the relationships between both EE (Effort Expectancy) and FC with BI. However, qualifications do not significantly moderate the relationships between latent variables and BI. The multigroup analysis highlights a significant divergence in the influence of PE on BI between groups with no experience and experienced people. Additionally, the study shows that the higher-income group displays a lower coefficient of FC towards BI, potentially due to their pre-existing knowledge base. The findings from this study assist decision-makers by providing insights into public attitudes towards AVs by revealing the key factors influencing public acceptance. [ABSTRACT FROM AUTHOR]

• Autonomous driving acceptance is confirmed and improved using the Kano model. • Twenty-one acceptance elements and four driver characteristics are defined. • Females show indifference toward ease of learning and purchase/maintenance cost. • As age increases, a conservative appearance is shown with many must-be attributes. • More driving experience resulted in more difficulty in ensuring acceptance. The development of advanced technology has revolutionized human life. In this regard, autonomous driving, a core technology currently being developed, is changing rapidly. In addition to improving technology, the acceptance of technology users must be secured. Most relevant studies conducted hitherto have involved evaluation using acceptance elements defined based on the technology acceptance model and the unified theory of acceptance and use of technology. In this study, 21 elements associated with the acceptance of autonomous driving are defined. The Kano model is used to classify the acceptance elements into five attributes and to propose guidelines for improving acceptance. Driver characteristics are classified based on four human factors, which are used to investigate differences in acceptance between groups. A Google survey and fieldwork were completed by 187 participants. Contrary to previous studies, no significant gender differences are observed in the current study. In terms of age, many obstacles are encountered in securing autonomous driving acceptance from the elderly driver group. Additionally, a more conservative tendency is indicated by people with more driving experience. The results of this study reveal important points for identifying elements that hinder future sustainability and commercialization of autonomous driving, thereby facilitating its further technological development. [ABSTRACT FROM AUTHOR]

• Enhancing the public transport service had only a marginal effect on ridership, except for rural areas where the impact was greater. • Leisure and service trips were more impacted by self-driving technology than commuting trips. • Walking and bicycling declined across all modelled scenarios. Self-driving technology may lead to a paradigm shift for the transport industry with shared cars available to every-one. However, this vision has increasingly been challenged as too optimistic and unsubstantiated. In this study we explore societal impacts of using this technology for both cars and public transport and investigate differences depending on geography and trip purpose. Four scenarios were designed through workshops with 130 transport experts, modelled using a conventional four-step model for Stockholm, Sweden and evaluated in terms of changes to mode choice, number of trips and person kilometres. We find larger increases for non-commuting trips, i.e. service and leisure trips, than for commuting trips, questioning the view of the 'productive work trip' as self-driving technology's main impact on society. As these trips are primarily made outside of rush hours, this may lead to a changed transport system. Geographic differences are substantial and heavily dependent on the cost model for car alternatives, even indicating a reduction in car travel in rural areas if private ownership would be replaced by shared cars. Furthermore, walking and cycling levels decreased in all scenarios while enhancing public transport using self-driving technology had a limited impact on ridership. These results show that the impacts of self-driving technology may have varied societal impacts even within a region and may lead to increased car travel, especially off-peak. These conclusions stress the need for policies that are sensitive to both geography and time. [ABSTRACT FROM AUTHOR]

• The Technology Acceptance Model of Autonomous vehicles for Tourism purposes (TAMAT) has been developed. • Openness to Tourism Usage (OTU) has a positive impact on Perceived Ease of Use (PEOU), Perceived Usefulness (PU), and thus on the Intention to Use (ITU). • In Unusual Surroundings (UNS) tourists are more open to using self-driving cars, and UNS strengthens the positive relationship between Openness to Tourism Usage (OTU) and Perceived Usefulness (PU). • Adherence to Conventional Use negatively influences ITU and dampens the positive impacts of OTU and UNS on ITU. Autonomous vehicles are expected to shape mobility and tourism. This paper introduces an extension to the TAM to better understand the adoption of self-driving cars for tourism purposes. The new model (TAMAT) confirms some under-explored impacts of tourism-related variables, such as Openness to Tourism Usage and Unusual Surroundings, and the Adherence to Conventional Use on the Intention to Use self-driving cars. The research is based on online data collection (n = 646) and applies Covariance-Based Structural Equation Modelling. Findings indicate that the opportunity of using self-driving cars for tourism and unusual environments has a positive impact, while adherence to conventional car use has a negative impact on the intention to use self-driving cars. [ABSTRACT FROM AUTHOR]

Rapid advancements in deep learning have led to many recent breakthroughs. While deep learning models achieve superior performance, often statistically better than humans, their adoption into safety-critical settings, such as healthcare or self-driving cars is hindered by their inability to provide safety guarantees or to expose the inner workings of the model in a human understandable form. We present MoËT , a novel model based on Mixture of Experts, consisting of decision tree experts and a generalized linear model gating function. Thanks to such gating function the model is more expressive than the standard decision tree. To support non-differentiable decision trees as experts, we formulate a novel training procedure. In addition, we introduce a hard thresholding version, MoËT h , in which predictions are made solely by a single expert chosen via the gating function. Thanks to that property, MoËT h allows each prediction to be easily decomposed into a set of logical rules in a form which can be easily verified. While MoËT is a general use model, we illustrate its power in the reinforcement learning setting. By training MoËT models using an imitation learning procedure on deep RL agents we outperform the previous state-of-the-art technique based on decision trees while preserving the verifiability of the models. Moreover, we show that MoËT can also be used in real-world supervised problems on which it outperforms other verifiable machine learning models. [ABSTRACT FROM AUTHOR]

Several deep learning algorithms are currently focused on object detection in adverse weather scenarios for autonomous driving systems. However, these algorithms face challenges in real-time scenarios, leading to a reduction in detection accuracy. To tackle these issues, this paper introduces a lightweight object detection model named Self Driving Cars You Only Look Once Adverse Weather Detection Network (SD-YOLO-AWDNet), derived from enhancements to the YOLOv5 algorithm. The model incorporates four progressive improvement levels within the YOLOv5 framework. This includes integrating C3Ghost and GhostConv modules in the backbone to enhance detection speed by reducing computational overhead during feature extraction. To address potential accuracy issues arising from these modules, Depthwise-Separable Dilated Convolutions (DSDC) are introduced, striking a balance between accuracy and parameter reduction. The model further incorporates a Coordinate Attention (CA) module in the GhostBottleneck to enhance feature extraction and eliminate unnecessary features, improving precision in object detection. Additionally, a novel "Focal Distribution Loss" replaces CIoU Loss, accelerating bounding box regression and loss reduction. Test dataset experiments demonstrate that SD-YOLO-AWDNet outperforms YOLOv5 with a 54% decrease in FLOPs, a 52.53% decrease in model parameters, a 2.24% increase in mAP, and a threefold improvement in detection speed. [ABSTRACT FROM AUTHOR]

Reducing traffic congestion and increasing passenger safety are important objectives for emerging automated transportation systems. Autonomous intersection management systems (AIMS) enable large scale optimization of vehicle trajectories with connected and autonomous vehicles (CAVs). We propose a novel approach for computing the fastest waypoint trajectory in intersections using graph search in a discretized space-time graph that produces collision-free paths with variable vehicle speeds that comply with traffic rules and vehicle dynamical constraints. To assist our planner algorithm in decentralized scenarios, we also propose a multi-agent protocol architecture for vehicle coordination for trajectory planning using a vehicle-to-vehicle (V2V) network. The trajectories generated allow a much higher evacuation rate and congestion threshold, with lower O (N) algorithm runtime compared to the state of the art conflict detection graph platoon path planning method, even for large scenarios with vehicle arrival rate of 1 / s and thousands of vehicles. [ABSTRACT FROM AUTHOR]

We introduce an experimental methodology designed to detect the position-vectors of omnidirectional targets located in a circular environment. This is achieved using multiple channels of probe chaotic waves generated by an asymmetric coupling semiconductor lasers network (ACSLN). The probe waves' key features of time–space uncorrelation and broad bandwidth are integral to this process. These features enable us to accurately measure position-vectors for a multitude of omnidirectional targets. This is accomplished by correlating the multiple channels of the probe waves with their respective reference waves. Our experimental outcomes indicate impressively low relative errors in the position-vector detection of omnidirectional targets, not amounting to more than 0.4%. By fine-tuning the coupling strength, we were able to achieve ranging-resolutions for the omnidirectional targets as high as 4.63 mm. While a single target within a distance of no more than 4.3 m has minimal impact on the resolution of ranging measurements, it significantly influenced both the relative error and the peak sidelobe level. Our successful detection of omnidirectional targets' position-vectors with the ACSLN opens up fascinating prospects for potential applications; notably in the fields of driverless cars and object tracking systems that require omnidirectional vision. These results may indeed revolutionize future developments in these areas. • We have developed an experimental configuration that enables precise ranging for omnidirectional multi-targets using the multiple channels of the probe chaotic wave from the laser network. • Our experimental outcomes validate that the detections for the position-vectors of the omnidirectional targets carry extremely low relative errors, not exceeding 0.4%. The ranging resolution to multi-targets can achieve an impressive highpoint of 4.7 mm. • The distance to arbitrary target minimally impacts the ranging resolution, but markedly impacts both the relative-error and peak sidelobe-level. [ABSTRACT FROM AUTHOR]

[Display omitted] • Black double-shell hollow nanoparticles (BDS-HNPs) is prepared as LiDAR-reflective materials. • Strategies of multiple interfaces and internal white shell result as a superb NIR reflectance. • With the hydrophilic nature, BDS-HNPs are easily formulated as eco-friendly hydrophilic paint. • BDS-HNPs are dually function as NIR reflective and black color exhibiting layer in monolayer. • Three types of LiDAR sensors are employed for practical recognition of BDS-HNPs-painted object. Novel LiDAR-detectable black double-shell hollow nanoparticles (BDS-HNPs) with internal white shell are successfully utilized as materials for autonomous vehicle paint for the first time. These BDS-HNPs are carefully designed to achieve excellent near-infrared (NIR) reflectance, blackness, hydrophilicity, and applicability as monolayer coatings. An emphasis is placed on the NIR reflectance by forming double-shell hollow morphologies embracing the internal white shell and multiple interfaces within the nanoparticles. Accordingly, the BDS-HNPs exhibit NIR reflectance of ca. 33.2, 36.9, and 40.9 R % at wavelengths of 793, 850, and 905 nm, respectively, comparable to NIR reflectance of the commercially available NIR-reflective bilayer dark-tone coating. For the practical LiDAR visualization, BDS-HNPs mixed with hydrophilic varnish are spray-coated onto the various objects. As a result, the BDS-HNPs-painted objects are clearly recognized by three different types of LiDAR sensors (robot, rotating, and MEMs mirror) under various conditions of inside and outside. These results clearly demonstrate the great potential of BDS-HNPs as a new type of LiDAR-detectable black material for future autonomous driving environments. [ABSTRACT FROM AUTHOR]

• Children and adults waiting longer to enter the roadway when vehicles decelerated abruptly than gradually. • When vehicles decelerated gradually, both children and adults in the early eHMI onset condition entered the roadway well before vehicles came to a stop. • Children and adults in the late eHMI onset condition waited to enter the roadway until vehicles came to a stop. • Children were heavily influenced by the late eHMI signal, crossing later when the eHMI signal appeared late and the vehicle decelerated either gradually or abruptly compared to the control condition. • Without an eHMI signal, children show a concerning overreliance on gradual vehicle deceleration to judge yielding intent. How autonomous vehicles (AVs) communicate their intentions to vulnerable road users (e.g., pedestrians) is a concern given the rapid growth and adoption of this technology. At present, little is known about how children respond to external Human Machine Interface (eHMI) signals from AVs. The current study examined how adults and children respond to the combination of explicit (eHMI signals) and implicit information (vehicle deceleration) to guide their road-crossing decisions. Children (8- to 12-year-olds) and adults made decisions about when to cross in front of a driverless car in an immersive virtual environment. The car sometimes stopped, either abruptly or gradually (manipulated within subjects), to allow participants to cross. When yielding, the car communicated its intent via a dome light that changed from red to green and varied in its timing onset (manipulated between subjects): early eHMI onset, late eHMI onset, or control (no eHMI). As expected, we found that both children and adults waited longer to enter the roadway when vehicles decelerated abruptly than gradually. However, adults responded to the early eHMI signal by crossing sooner when the cars decelerated either gradually or abruptly compared to the control condition. Children were heavily influenced by the late eHMI signal, crossing later when the eHMI signal appeared late and the vehicle decelerated either gradually or abruptly compared to the control condition. Unlike adults, children in the control condition behaved similarly to children in the early eHMI condition by crossing before the yielding vehicle came to a stop. Together, these findings suggest that early eHMI onset may lead to riskier behavior (initiating crossing well before a gradually decelerating vehicle comes to a stop), whereas late eHMI onset may lead to safer behavior (waiting for the eHMI signal to appear before initiating crossing). Without an eHMI signal, children show a concerning overreliance on gradual vehicle deceleration to judge yielding intent. [ABSTRACT FROM AUTHOR]

The evolution of Autonomous Vehicles (AVs) has blurred the distinction between drivers and passengers, resulting in increased demand for in-car entertainment. Simultaneously, social networking platforms are evolving to adopt AR (Augmented Reality)/VR (Virtual Reality)-centric metaverses. Self-driving cars will free up the drivers and passengers to engage in leisure activities such as watching VR videos and playing games while travelling. Before driverless cars can become a "third living space", open challenges confront the provision of immersive content (e.g., VR videos and gaming) with a high level of Quality of Experience (QoE). Current autonomous vehicular network technologies focus on navigation and safety and rely on ubiquitous high-speed 5G/6G network connectivity to provide a seamless streaming experience in AVs. The compounded risks of cyber- and motion-sickness are identified as barriers to the enjoyment of in-car VR entertainment technologies. The influence of AV car networks in different traffic conditions is identified as a key factor to minimize a person's risk of sickness and increase overall QoE. However, assessing the QoE of immersive media consumption in AV to its full potential is a difficult task that requires proper standard guidelines and directions. Existing surveys focus only on a specific metaverse or transportation system. These surveys lack a comprehensive view of the entire immersive media consumption process, QoE factors and aspects, and QoE standards for assessing immersive content in vehicular networks. To address these issues, we present an overview of the research landscape in the context of QoE and network standards that highlights the need for experimental studies to inform and consolidate the methodologies and recommendations for standardization of immersive content QoE assessment in AVs. Research and user studies have not yet yielded standards or recommendations describing assessment methodologies to evaluate immersive systems, such as exist for traditional video and images. A framework is outlined to provide a comprehensive understanding of the QoE influencing factors and aspects that will assist in adapting, augmenting, and producing standards for immersive entertainment in autonomous vehicles. • The convergence of Autonomous Vehicles (AVs) and immersive media in "Moving Metaverse" brings in new in-car entertainment opportunities like VR videos and gaming during travel. • Challenges confront AVs from becoming a "third living space," including delivering high-quality immersive content (e.g., VR videos, gaming) due to network connectivity, motion sickness, and cybersickness. • To ensure user satisfaction when integrating immersive experiences into AVs, a thorough understanding of QoE factors such as cybersickness, presence, and immersion is required. • The survey identifies gaps in existing research, particularly in the standardization of QoE assessment methodologies for immersive content consumption in AVs and vehicular networks. • The survey helps to bridge the gap between industry advancements and academic research by providing a roadmap for future research goals and QoE standardization activities in the context of the "Moving Metaverse". [ABSTRACT FROM AUTHOR]

• Developed framework for creating AV's ambiguousdriving scenarios. • Created 28 ambiguous driving scenarios andexamined 548 AV users' perspectives. • Scenarios grouped by AV users' moral, ethical,legal, utility, and safety perceptions. • User perception varies across scenarios, safetyranking highest. • Study results offer valuable insights for AVpolicy and algorithm advancement. Human drivers are gradually being replaced by highly automated driving systems, and this trend is expected to persist. The response of autonomous vehicles to Ambiguous Driving Scenarios (ADS) is crucial for legal and safety reasons. Our research focuses on establishing a robust framework for developing ADS in autonomous vehicles and classifying them based on AV user perceptions. To achieve this, we conducted extensive literature reviews, in-depth interviews with industry experts, a comprehensive questionnaire survey, and factor analysis. We created 28 diverse ambiguous driving scenarios and examined 548 AV users' perspectives on moral, ethical, legal, utility, and safety aspects. Based on the results, we grouped ADS, with all of them having the highest user perception of safety. We classified these scenarios where autonomous vehicles yield to others as moral, bottleneck scenarios as ethical, cross-over scenarios as legal, and scenarios where vehicles come to a halt as utility-related. Additionally, this study is expected to make a valuable contribution to the field of self-driving cars by presenting new perspectives on policy and algorithm development, aiming to improve the safety and convenience of autonomous driving. [ABSTRACT FROM AUTHOR]

In self-driving cars, adverse weather (e.g., fog, rain, snow, and cloud) or occlusion scenarios result in domain shift being unavoidable in object detection. Researchers have recently proposed Domain Adaptive Object Detection (DAOD), i.e., aligning the source and target domains at the image and instance levels distribution by utilizing the Unsupervised Domain Adaptation (UDA) method. However, the semantic correlation information is ignored leading to the effect of aligning not good, and low detection accuracy of objects in adverse weather or occlusion scenarios. Here, we propose a Dual Semantic Correlation Alignment (DSCA) method for DAOD to address the problem. The core idea behind DSCA is to make full use of semantic correlation information including context correlation semantic information and class correlation semantic information to align object semantic information in source and target domains, which supplement and enhance the missing information for target domains. It consists of a two-level semantic alignment: (1) context correlation semantic alignment is developed to obtain the context correlation semantic information of the object to align context semantic information at the image level; (2) class correlation semantic alignment is proposed to obtain the class correlation semantic information of the object to align class semantic information at the instance level. The two-level semantic alignment can effectively decrease negative transfer and complete object information to improve the detection accuracy of objects in different domains. Experiments on four challenging benchmarks show that our proposed DSCA method outperforms state-of-the-art DAOD methods. • We propose a Dual Semantic Correlation Alignment (DSCA) method for domain adaptation object detection. The DSCA takes full advantage of context and class correlation semantics information to align object semantic information in source and target domains. • The Context Correlation Semantic Alignment (COCSA) including the Local Context Correlation Semantic (LCCS) and Global Context Correlation Semantic (GCCS) modules is proposed to align the context semantic information at the image level. • The Class Correlation Semantic Alignment (CLCSA) comprising the Class Semantic Decoupling (CSD) and Dynamic Adaptation Graph Convolutional Network (DA-GCN) modules is proposed to align class semantic information at the instance level. • We conduct extensive experiments showing that the proposed DSCA comprising the COCSA and CLCSA modules in DAOD outperforms the state-of-the-art on four challenging benchmarks. [ABSTRACT FROM AUTHOR]

With the emerging deployment of deep neural networks, such as in mobile devices and autonomous cars, there is a growing demand for neural architecture search (NAS) to automatically design powerful network architectures. It is more reasonable to formulate NAS as a multi-objective optimization problem. In addition to prediction performance, multi-objective NAS (MONAS) problems take into account other criteria like the number of parameters and inference latency. Multi-objective evolutionary algorithms (MOEAs) are the preferred approach for tackling MONAS due to their effectiveness in dealing with multi-objective optimization problems. Recently, local search-based NAS algorithms have demonstrated their efficiency over MOEAs for MONAS problems. However, their performance has been only verified on bi-objective NAS problems. In this article, we propose a local search algorithm for multi-objective NAS (LOMONAS), an efficient local search framework for solving not only bi-objective NAS problems but also NAS problems having more than two objectives. We additionally present a parameter-less version of LOMONAS, namely IMS-LOMONAS, by combining LOMONAS with the Interleaved Multi-start Scheme (IMS) to help NAS practitioners avoid manual control parameter settings. Experimental results from a series of benchmark problems in the CEC'23 Competition demonstrate the competitiveness of LOMONAS and IMS-LOMONAS compared to MOEAs in tackling MONAS within both small-scale and large-scale search spaces. Source code is available at: https://github.com/ELO-Lab/IMS-LOMONAS. [ABSTRACT FROM AUTHOR]

• Participants responded to three actual misconceptions about AVs and attitudes toward AVs. • One-third believed AVs are commercially available. • Participants who hold more misconceptions about AVs may be more positive toward AVs. • Dispelling public misconceptions about AVs is necessary. Automated vehicle (AV) technology is likely to influence transportation, mobility, and society dramatically. The year 2020 was a horizon year for the AV, as manufacturers expected commercial AVs to be available to the general market. However, we experienced one cycle of hyperbole for these "self-driving" cars, which are still unavailable to consumers. Meanwhile, many persistent beliefs about this technology are factual or arguable misconceptions. However, the public attitude literature rarely examines public misconceptions of AVs. Thus, we explored the prevalence of three misconceptions: "AVs are already available in the market," "AVs do not need to be driven manually at all," and "Mature business models for AVs have been established." We investigated these misconceptions' correlations with several cognitive (i.e., benefit and risk perceptions), affective (i.e., positive and negative affect), and behavioral components (i.e., behavioral intention and willingness to pay) of attitudes and trust in AVs. Our online survey in China (N = 1,026) indicated that more than 70% of participants hold one or more of the three misconceptions, with one-third believing that AVs are already available in the market. Furthermore, participants believing one or more of the three misconceptions were more positive regarding specific attitudinal factors (e.g., those who believed that AVs are already available reported greater behavioral intention to use and willingness to pay for AVs than those who rejected this misbelief). This finding indicates that people who are more wrong about AVs might be more positive toward AVs. We need effective and accurate public communication to dispel public misconceptions about AVs and build rational expectations. [ABSTRACT FROM AUTHOR]

These days, due to the emergence of Industry 4.0, it is possible to notice a growing interest among companies in automating their processes. Autonomous driving technologies are generating increasing interest in carrying out material handling operations, being clear that these technologies allow companies to have greater flexibility, higher accuracy, lower costs, and higher productivity. What is not clear is how to implement these autonomous driving vehicles, which raises questions such as: What are the requirements to consider when using autonomous transport to move materials in the shopfloor? What are steps must be taken to introduce autonomous driving technologies in an industrial environment? This article, which describes a case study carried out in an automotive company, provides the answers to these questions, as well as presents a methodology to implement Autonomous Guided Vehicles (AGVs), being a relevant contribution to the understanding and implementation of AGVs in the industrial context. During the case study, the number of AGVs required was estimated (by analytical method and using SIMIO simulation software), and an economic feasibility study of the project was also performed. [ABSTRACT FROM AUTHOR]

Based on historical records, driving in hazardous weather conditions is one of the most serious causes that lead to fatal accidents on roads in general and in United Arab Emirates (UAE) highways in particular. One solution to improve road safety is to equip vehicles and infrastructure with connected and smart devices and convert them into autonomous vehicles. Before deploying a concrete solution to the field, it must be validated by simulation, and more specifically by agent-based simulation. In this paper, we propose to implement the Reaction Time-Based Collaborative Velocity Control (RT-CVC) model that was implemented in autonomous cars into an agent-based simulator. This model is compared to the Intelligent Driver Model (IDM), which is one of the standard longitudinal driving behaviors in simulation environments. The experimental results show that RT-CVC generates traffic flows with fewer vehicle collisions and shorter travel times. This positive analysis is balanced by the fact that RT-CVC is designed for autonomous cars, and IDM is designed to model human-drive decisions. Using RT-CVC for modeling a human driver may be counter productive in simulation experiments. [ABSTRACT FROM AUTHOR]