Semi-Direct Monocular Visual-Inertial Odometry Using Point and Line Features for IoV

The precise measuring of vehicle location has been a critical task in enhancing the autonomous driving in terms of intelligent decision making and safe transportation. Internet of Vehicles ( IoV ) is an important infrastructure in support of autonomous driving, allowing real-time road information exchanging and sharing for localizing vehicles. Global positioning System ( GPS ) is widely used in the traditional IoV system. GPS is unable to meet the key application requirements of autonomous driving due to meter level error and signal deterioration. In this article, we propose a novel solution, named Semi-Direct Monocular Visual-Inertial Odometry using Point and Line Features ( SDMPL-VIO ) for precise vehicle localization. Our SDMPL-VIO model takes advantage of a low-cost Inertial Measurement Unit ( IMU ) and monocular camera, using them as the sensor to acquire the surrounding environmental information. Visual-Inertial Odometry ( VIO ), taking into account both point and line features, is proposed, which is able to deal with both weak texture and dynamic environment. We use a semi-direct method to deal with keyframes and non-keyframes, respectively. Dual sliding window mechanisms can effectively fuse point-line and IMU information. To evaluate our SDMPL-VIO system model, we conduct extensive experiments on both an indoor dataset (i.e., EuRoC) and an outdoor dataset (i.e., KITTI) from the real-world applications, respectively. The experimental results show that the accuracy of SDMPL-VIO proposed by us is better than the mainstream VIO system at present. Especially in the weak texture of the datasets, fast-moving datasets, and other challenging datasets, SDMPL-VIO has a relatively high robustness.

PRIMA: general and precise neural network certification via scalable convex hull approximations

Formal verification of neural networks is critical for their safe adoption in real-world applications. However, designing a precise and scalable verifier which can handle different activation functions, realistic network architectures and relevant specifications remains an open and difficult challenge. In this paper, we take a major step forward in addressing this challenge and present a new verification framework, called PRIMA. PRIMA is both (i) general: it handles any non-linear activation function, and (ii) precise: it computes precise convex abstractions involving multiple neurons via novel convex hull approximation algorithms that leverage concepts from computational geometry. The algorithms have polynomial complexity, yield fewer constraints, and minimize precision loss. We evaluate the effectiveness of PRIMA on a variety of challenging tasks from prior work. Our results show that PRIMA is significantly more precise than the state-of-the-art, verifying robustness to input perturbations for up to 20%, 30%, and 34% more images than existing work on ReLU-, Sigmoid-, and Tanh-based networks, respectively. Further, PRIMA enables, for the first time, the precise verification of a realistic neural network for autonomous driving within a few minutes.

A TDMA-Based MAC Protocol for Mitigating Mobility-Caused Packet Collisions in Vehicular Ad Hoc Networks

An autonomous driving environment poses a very stringent requirement for the timely delivery of safety messages in vehicular ad hoc networks (VANETs). Time division multiple access (TDMA)-based medium access control (MAC) protocols are considered a promising solution because of their time-bound message delivery. However, in the event of mobility-caused packet collisions, they may experience an unpredicted and extended delay in delivering messages, which can cause catastrophic accidents. To solve this problem, a distributed TDMA-based MAC protocol with mobility-caused collision mitigation (MCCM-MAC) is presented in this paper. The protocol uses a novel mechanism to detect merging collisions and mitigates them by avoiding subsequent access collisions. One vehicle in the merging collisions retains the time slot, and the others release the slot. The common neighboring vehicles can timely suggest a suitable new time slot for the vacating vehicles, which can avoid access collisions between their packet transmissions. A tie-breakup mechanism is employed to avoid further access collisions. Simulation results show that the proposed protocol reduces packet loss more than the existing methods. Consequently, the average delay between the successfully delivered periodic messages is also reduced.

Pyramid Bayesian Method for Model Uncertainty Evaluation of Semantic Segmentation in Autonomous Driving

With the fast-developing deep learning models in the field of autonomous driving, the research on the uncertainty estimation of deep learning models has also prevailed. Herein, a pyramid Bayesian deep learning method is proposed for the model uncertainty evaluation of semantic segmentation. Semantic segmentation is one of the most important perception problems in understanding visual scene, which is critical for autonomous driving. This study to optimize Bayesian SegNet for uncertainty evaluation. This paper first simplifies the network structure of Bayesian SegNet by reducing the number of MC-Dropout layer and then introduces the pyramid pooling module to improve the performance of Bayesian SegNet. mIoU and mPAvPU are used as evaluation matrics to test the proposed method on the public Cityscapes dataset. The experimental results show that the proposed method improves the sampling effect of the Bayesian SegNet, shortens the sampling time, and improves the network performance.

A Novel Vehicle Detection Framework Based on Parallel Vision

Autonomous driving has become a prevalent research topic in recent years, arousing the attention of many academic universities and commercial companies. As human drivers rely on visual information to discern road conditions and make driving decisions, autonomous driving calls for vision systems such as vehicle detection models. These vision models require a large amount of labeled data while collecting and annotating the real traffic data are time-consuming and costly. Therefore, we present a novel vehicle detection framework based on the parallel vision to tackle the above issue, using the specially designed virtual data to help train the vehicle detection model. We also propose a method to construct large-scale artificial scenes and generate the virtual data for the vision-based autonomous driving schemes. Experimental results verify the effectiveness of our proposed framework, demonstrating that the combination of virtual and real data has better performance for training the vehicle detection model than the only use of real data.

Research on Integrated Decision Control Algorithm for Autonomous Vehicles Under Multi-Task Hybrid Constraints in Intelligent Transportation Scenarios

In order to improve the decision-making and control effect of autonomous vehicles, in this paper, combined with literature research and process analysis, the control algorithm of autopilot vehicle is analyzed, and the driving process is analyzed combined with the flow method. In order to improve the effect of autonomous driving, with the support of improved algorithms, an integrated decision-making control system for autonomous vehicles under multi-task constraints in intelligent traffic scenarios is constructed, and system performance is improved by simulating autonomous driving decisions in a variety of complex situations. Moreover, this paper designs the road driving model according to actual needs, sets the functional modules of the entire system, and build the overall framework of the system. Finally, in order to study the integrated decision-making effect of this system, this paper conducts test research by designing a simulation test method. From the simulation test results, it can be seen that the intelligent decision-making system for autonomous vehicles constructed in this paper has certain effects.

Mental workLoad Accumulation Effect of Mobile Phone Distraction in L2 Autopilot Mode

As self-driving vehicles become more common, there is a need for precise measurement and definition of when and in what ways a driver can use a mobile phone in autonomous driving mode, for how long it can be used, the complexity of the call content, and the accumulated psychological load. This study uses a 2 (driving mode) * 2 (call content complexity) * 6 (driving phase) three-factor mixed experimental design to investigate the effect of these factors on the driver's psychological load by measuring the driver's performance on peripheral visual detection tasks, pupil diameter, and EEG components in various brain regions in the alpha band. The results showed that drivers' mental load levels converge between manual and automatic driving modes as the duration of driving increases, regardless of the level of complexity of the mobile phone conversation. This suggests that mobile phone conversations can also disrupt the driver's cognitive resource balance in automatic driving mode, as it increases mental load while also impairing the normal functioning of brain functions such as cognitive control, problem solving, and judgment, thereby compromising driving safety.

A Comprehensive Survey on Radio Frequency (RF) Fingerprinting: Traditional Approaches, Deep Learning, and Open Challenges

Fifth generation (5G) networks and beyond envisions massive Internet of Things (IoT) rollout to support disruptive applications such as extended reality (XR), augmented/virtual reality (AR/VR), industrial automation, autonomous driving, and smart everything which brings together massive and diverse IoT devices occupying the radio frequency (RF) spectrum. Along with spectrum crunch and throughput challenges, such a massive scale of wireless devices exposes unprecedented threat surfaces. RF fingerprinting is heralded as a candidate technology that can be combined with cryptographic and zero-trust security measures to ensure data privacy, confidentiality, and integrity in wireless networks. Motivated by the relevance of this subject in the future communication networks, in this work, we present a comprehensive survey of RF fingerprinting approaches ranging from a traditional view to the most recent deep learning (DL) based algorithms. Existing surveys have mostly focused on a constrained presentation of the wireless fingerprinting approaches, however, many aspects remain untold. In this work, however, we mitigate this by addressing every aspect - background on signal intelligence (SIGINT), applications, relevant DL algorithms, systematic literature review of RF fingerprinting techniques spanning the past two decades, discussion on datasets, and potential research avenues - necessary to elucidate this topic to the reader in an encyclopedic manner.

Point Cloud Semantic Segmentation with Cross-Correction Features

Real time large scale point cloud segmentation is an important but challenging task for practical application like autonomous driving. Existing real time methods have achieved acceptance performance by aggregating local information. However, most of them only exploit local spatial information or local semantic information dependently, few considering the complementarity of both. In this paper, we propose a model named Spatial-Semantic Incorporation Network (SSI-Net) for real time large scale point cloud segmentation. A Spatial-Semantic Cross-correction (SSC) module is introduced in SSI-Net as a basic unit. High quality contextual features can be learned through SSC by correct and update semantic features using spatial cues, and vice verse. Adopting the plug-and-play SSC module, we design SSI-Net as an encoder-decoder architecture. To ensure efficiency, it also adopts a random sample based hierarchical network structure. Extensive experiments on several prevalent datasets demonstrate that our method can achieve state-of-the-art performance.

Data collection protocols for VANETs: a survey

We live in the era of Intelligent Transport Systems (ITS), which is an extension of Vehicular AdHoc Networks (VANETs). In VANETs, vehicles act as nodes connected with each other and sometimes with a public station. Vehicles continuously exchange and collect information to provide innovative transportation services; for example, traffic management, navigation, autonomous driving, and the generation of alerts. However, VANETs are extremely challenging for data collection, due to their high mobility and dynamic network topologies that cause frequent link disruptions and make path discovery difficult. In this survey, various state-of-the-art data collection protocols for VANETs are discussed, based on three broad categories, i.e., delay-tolerant, best-effort, and real-time protocols. A taxonomy is designed for data collection protocols for VANETs that is essential to add precision and ease of understandability. A detailed comparative analysis among various data collection protocols is provided to highlight their functionalities and features. Protocols are evaluated based on three parametric phases. First, protocols investigation based on six necessary parameters, including delivery and drop ratio, efficiency, and recovery strategy. Second, a 4-D functional framework is designed to fit most data collection protocols for quick classification and mobility model identification, thus eradicating the need to read extensive literature. In the last, in-depth categorical mapping is performed to deep dive for better and targeted interpretation. In addition, some open research challenges for ITS and VANETs are discussed to highlight research gaps. Our work can thus be employed as a quick guide for researchers to identify the technical relevance of data collection protocols of VANETs.