The emergence of autonomous driving represents a pivotal milestone in the evolution of the transportation system, integrating seamlessly into the daily lives of individuals due to its array of advantages over conventional vehicles. However, self-driving cars pose numerous challenges contributing to accidents and injuries annually. This paper aims to comprehensively examine the limitations inherent in autonomous driving and their consequential impact on accidents and collisions. Using data from the DMV, NMVCCS, and NHTSA, the paper reveals the key factors behind self-driving car accidents. It delves into prevalent limitations faced by self-driving cars, encompassing issues like adverse weather conditions, susceptibility to hacking, data security concerns, technological efficacy, testing and validation intricacies, information handling, and connectivity glitches. By meticulously analyzing reported accidents involving self-driving cars during the period spanning 2019 to 2022, the research evaluates statistical data pertaining to fatalities and injuries across diverse accident classifications. Additionally, the paper delves into the ethical and regulatory dimensions associated with autonomous driving, accentuating the legal complexities that arise from accidents involving self-driving vehicles. This review assists researchers and professionals by identifying current autonomous driving limitations and offering insights for safer adoption. Addressing these limitations through research can transform transportation systems for the better.
13 June 2024 in IEEE Transactions on Vehicular Technology
The modern Intelligent Vehicle (IV) is a complex technological marvel that heavily relies on the Controller Area Network (CAN) bus system to enable seamless communication among different electronic control units (ECUs). However, the CAN bus system lacks security mechanisms for authentication and authorization, leaving it vulnerable to various attacks. Malicious actors can freely broadcast CAN messages without protection, making the system susceptible to DoS, Fuzzing, and Spoofing attacks. Therefore, it is crucial to devise methods to safeguard modern vehicles from such threats. In this research paper, we introduce HybridSecNet, A hybrid two-step LSTM-CNN Model for Intrusion Detection, a deep learning-based architecture specifically designed to bolster in-vehicle security on Controller Area Networks (CAN). HybridSecNet comprises two stages of classification: the first stage employs long short-term memory (LSTM) to categorize input data as either normal or attacked, and the second stage further classifies the attacks into specific types using Convolutional Neural Networks (CNN). This two-step approach significantly enhances classification accuracy and reliability, yielding remarkable results with accuracy, precision, recall, and an F1-score of approximately 99.5% for CAN bus network attacks. Comparative analyses with existing single-step models underscore the superiority of our proposed model, demonstrating its potential to revolutionize in-vehicle security in the realm of modern intelligent vehicles.
8 June 2022 in IEEE Geoscience and Remote Sensing Letters
Object Detection in low-light aerial images is a challenging problem due to considerable variation in brightness and varying contrast. Deep Learning-based approaches have recently demonstrated great promise in image enhancement. Many existing neural networks used for image quality enhancement first encode the input into low-resolution representations and then decode these representations back to a higher resolution for the contextual information. However, this method leads to the loss of semantic content. Recent research has demonstrated the advantage of maintaining high-resolution information along with lower resolution representations, which maintains image features throughout the network. In this paper, we propose a novel architecture named RNet for low-light image enhancement of aerial images. The proposed network contains multi-resolution branches for better understanding of different levels of local and global context through different streams. The performance of RNet is evaluated on a recent synthetic dataset. We also present a comprehensive evaluation with a representative set of state-of-the-art enhancement techniques and neural net architectures.
The field of autonomous driving research has made significant strides towards achieving full automation, endowing vehicles with self-awareness and independent decision-making. However, integrating automation into vehicular operations presents formidable challenges, especially as these vehicles must seamlessly navigate public roads alongside other cars and pedestrians. An intriguing yet relatively underexplored domain within autonomous driving is overtaking. Overtaking involves a dynamic interplay of complex tasks, including precise steering and speed control, rendering it one of the most intricate operations for implementing augmented intelligence driving technologies. Surprisingly, the overtaking of autonomous vehicles remains largely uncharted territory in the context of augmented intelligence for autonomous systems. This void in knowledge beckons researchers to embark on explorations and investigations in this nascent field. Our review paper systematically synthesises overtaking methodologies hinging on computer vision techniques tailored for augmented intelligence autonomous driving scenarios in response to this pressing need. Our analysis encompasses an array of domains central to overtaking in augmented intelligence autonomous vehicles, encompassing Object Detection, Lane/Line Detection, Depth Estimation, Obstacle Detection, Segmentation, and Pedestrian Detection. We meticulously analyze each domain using well-established Multimodal datasets. We assess different models’ performance across various parameters by employing graphical structures, enabling visual comparative analyses. In object detection, YOLOv4 achieves a top performance with 0.90 mAP on the BDD100K dataset. For lane detection, CLRNET excels with the highest F1 score of around 0.96 on the LLAMAS dataset. ViT-Adapter-L leads in segmentation tasks, boasting an impressive mIoU score of 83 on Cityscapes. The Hierarchical Model achieves a superior mAP of 0.90 in road sign detection on the Tsinghua-Tencent Dataset. Steering angle computation sees InterFuser as the standout, achieving the highest driving score of approximately 74.0. This paper’s primary contributions include a comprehensive assessment of diverse models for each Multimodal dataset, aiding future research in this evolving domain.
5 October 2023 in IEEE Transactions on Consumer Electronics
Federated learning has shown exceptional usability in diverse domains, such as IoT, healthcare, FinTech, Insurance Sector and other Industries and Technologies. In numerous emerging VANET applications, integrating federated learning and vertically partitioned (VP) systems has yielded advantageous outcomes. There is still much to discover in the field of traffic management and intelligent mobility. The adaptation of federated learning in ITS makes it safer, efficient and enhances vehicular mobility. The majority of federated learning techniques for vertically partitioned data are synchronous in nature. The development of asynchronous training algorithms for VP data is crucial while maintaining the privacy of the data to increase efficiency. This paper presents a Privacy-Preserving Asynchronous Federated Learning and Vertical partitioning based Algorithm that reduces the duration of idling at red lights and improves the fuel consumption of ITS. Additionally, we have introduced a technique that enhances the passing-vehicle ratio at the intersection and efficiently manages the traffic. Using a mathematical approach, we showed the fuel consumption dependency over time for traveling and fuel consumption. In the end, we evaluated our proposed model using various parameters, which shows our model’s efficiency. Our suggested model outperforms traditional vehicle passing per unit time ratio approaches. The ratio value provided by the traditional model is 0.88; however, the ratio value obtained by our proposed model sets a new benchmark, which is 1.33
7 September 2022 in IEEE Transactions on Intelligent Transportation Systems
There has been a notable increase in the research and development of Vehicular Ad-hoc Networks (VANETs) to efficiently and safely manage large amounts of traffic. Such networks are, however, also prone to various cyber threats to data integrity, privacy, authentication, and network availability, and given the potential risk to life under the event of a malfunction and misinformation, it is important to provide security measures against such threats. This paper presents the Multi-branch Reconstruction Error (MbRE) Intrusion Detection System (IDS) for edge-based anomaly detection in VANETs for data integrity, network availability and user authentication-based misbehaviors without the need to train on them. Vehicular data is first sequenced and separated into three data branches -frequency (F) derived from the message timestamps, pseudo-identities (I), and the motion data (M) i.e. position and velocity. The proposed model comprises of three Convolutional Neural Networks (CNN)-based reconstruction models trained to reconstruct normal F-I-M vehicular behavior. The IDS classifies each branch of a sequence as 0/1 based on the reconstruction error threshold for the respective branch and, therefore, has the ability to detect 8 possible binary encoded behaviors for each sequence of vehicular data. These results are then used to find the overall behavior of each vehicle using carefully selected detection thresholds. MbRE is able to classify frequency, identity and motion-based behavior samples with an accuracy of 100%, 98.5-100%, and 95.4-100%, respectively, without the need to train on such behaviors. The study also emulates the IDS on Google Colaboratory and Jetson Nano to show its practicality in cloud and edge environments.