3D Vehicle Research Articles

Somersault Foraging and Elite Opposition-Based Learning Dung Beetle Optimization Algorithm

To tackle the shortcomings of the Dung Beetle Optimization (DBO) Algorithm, which include slow convergence speed, an imbalance between exploration and exploitation, and susceptibility to local optima, a Somersault Foraging and Elite Opposition-Based Learning Dung Beetle Optimization (SFEDBO) Algorithm is proposed. This algorithm utilizes an elite opposition-based learning strategy as the method for generating the initial population, resulting in a more diverse initial population. To address the imbalance between exploration and exploitation in the algorithm, an adaptive strategy is employed to dynamically adjust the number of dung beetles and eggs with each iteration of the population. Inspired by the Manta Ray Foraging Optimization (MRFO) algorithm, we utilize its somersault foraging strategy to perturb the position of the optimal individual, thereby enhancing the algorithm’s ability to escape from local optima. To verify the effectiveness of the proposed improvements, the SFEDBO algorithm is utilized to optimize 23 benchmark test functions. The results show that the SFEDBO algorithm achieves better solution accuracy and stability, outperforming the DBO algorithm in terms of optimization results on the test functions. Finally, the SFEDBO algorithm was applied to the practical application problems of pressure vessel design, tension/extension spring design, and 3D unmanned aerial vehicle (UAV) path planning, and better optimization results were obtained. The research shows that the SFEDBO algorithm proposed in this paper is applicable to actual optimization problems and has better performance.

Self-supervised 3D vehicle detection based on monocular images

The deep learning-based 3D object detection literature on monocular images has been dominated by methods that require supervision in the form of 3D bounding box annotations for training. However, obtaining sufficient 3D annotations is expensive, laborious and prone to introducing errors. To address this problem, we propose a monocular self-supervised approach towards 3D object detection relying solely on observed RGB data rather than 3D bounding boxes for training. We leverage differentiable rendering to apply visual alignment to depth maps, instance masks and point clouds for self-supervision. Furthermore, considering the complexity of autonomous driving scenes, we introduce a point cloud filter to reduce noise impact and design an automatic training set pruning strategy suitable for the self-supervised framework to further improve network performance. We provide detailed experiments on the KITTI benchmark and achieve competitive performance with existing self-supervised methods as well as some fully supervised methods.

Indirect frequency estimation by time-shifted accelerations subtraction: Generalization of the methodology and numerical application on a Warren truss bridge

In the context of Structural Health Monitoring (SHM) of transportation infrastructures, bridge eigen frequencies are commonly adopted as diagnostic indices. Under this regard, as a cost-effective identification methodology, indirect SHM, which consists in the use of sensors installed on running vehicles, is attracting attention. In particular, methods relying on the use of two or more on-board sensors may suppress the influence of track irregularity, which is a remarkable drawback for indirect SHM, and improve the identification accuracy. However, the combination characteristics of the two on-board sensors is still an open issue to be addressed. Also, in many studies, test bridges consist of simple girders, and there are few verifications on complex structures, such as truss bridges. The first contribution of this study is the analytical and generalized expression of the operational range defining the use of time-shifted signals subtraction operation for indirect frequency estimation. After the outcomes provided by this formulation were confirmed by the results obtained from a 2D explanatory example, the proposed procedure was further validated through three-dimensional train-track-bridge dynamic simulations. For this purpose, 3D FE model of a real existing Warren truss bridge, dynamically validated through experimental tests, and a 3D multi-body rail vehicle were adopted. From an engineering standpoint, a relevant result of this study found is that the coexistence of proper sensors onboard positioning and sufficient target bridge component excitation revealed to be crucial for the success of an indirect frequency identification procedure which is based on time-shifted accelerations subtraction.

A Review of Generative Models for 3D Vehicle Wheel Generation and Synthesis

Integrating deep learning methodologies is pivotal in shaping the continuous evolution of computer-aided design (CAD) and computer-aided engineering (CAE) systems. This review explores the integration of deep learning in CAD and CAE, particularly focusing on generative models for simulating 3D vehicle wheels. It highlights the challenges of traditional CAD/CAE, such as manual design and simulation limitations, and proposes deep learning, especially generative models, as a solution. The study aims to automate and enhance 3D vehicle wheel design, improve CAE simulations, predict mechanical characteristics, and optimize performance metrics. It employs deep learning architectures like variational autoencoders (VAEs), convolutional neural networks (CNNs), and generative adversarial networks (GANs) to learn from diverse 3D wheel designs and generate optimized solutions. The anticipated outcomes include more efficient design processes, improved simulation accuracy, and adaptable design solutions, facilitating the integration of deep learning models into existing CAD/CAE systems. This integration is expected to transform design and engineering practices by offering insights into the potential of these technologies.

An Optimal Hierarchical Control Strategy for 4WS-4WD Vehicles Using Nonlinear Model Predictive Control

Advanced driving algorithms, control strategies, and their optimization in self-driving vehicles in various scenarios are hotspots in current research; 4WS-4WD (four-wheel steering and four-wheel drive) is another hotspot in the study of new concept models; and the nonlinear dynamic characteristics of self-driving vehicles (AVs) are prominent in the fast cornering mode, which leads to a significant reduction in the accuracy and stability of trajectory tracking. Based on these research backgrounds, this paper proposes a control strategy optimization idea based on the 4WS4WD vehicle and its optimization model. The main content includes the establishment of a 3D vehicle model that takes into account vehicle load transfer and position change, and the establishment of a hierarchical control strategy based on the optimized NMPC and 4WS4WD models. The controller consists of two parts: an upper tracking controller based on the new vehicle model and NMPC, and a lower decoupled controller. The tracking control effect of the algorithmic control strategy based on the model and controller is validated in the high-speed serpentine motion mode and double-shift linear motion mode on the joint simulation platform of Car Sim and Simulink.

A Deep Learning-Based CAE Approach for Simulating 3D Vehicle Wheels Under Real-World Conditions

The implementation of deep learning (DL) in computer-aided engineering (CAE) can significantly improve the accuracy and efficiency of simulating 3D vehicle wheels under real-world conditions. While traditional CAE methods can be time-consuming and computationally expensive, DL can reduce simulation time and development cycles across all industries. This work explores the role of DL and AI in virtual manufacturing and CAE and investigates how they can be used to improve the accuracy and efficiency of simulations for 3D vehicle wheels. Deep learning models can learn the complex relationships between different wheel design parameters, such as tire load distribution, stress distribution, and fatigue life. Once trained, these models can be embedded into CAE software, allowing for faster and more accurate simulations of wheel performance. This interdisciplinary study uses various deep learning techniques, including convolutional neural networks (CNNs), generative adversarial networks (GANs), and recurrent neural networks (RNNs), to create a more efficient and accurate relationship between CAD modeling and CAE simulation. The research aims to leverage the potential of deep learning models to automate 3D CAD design, accurately predict CAE results, and provide in-depth explanations and verifications. The benefits of this research are expected to extend to the automotive industry's pursuit of more robust and resilient wheel designs. By streamlining the product development process from conceptual design to engineering performance evaluation, this study has the potential to revolutionize the automotive industry's product development cycle.

An Attention Mechanism Based AVOD Network for 3D Vehicle Detection

An Attention Mechanism Based AVOD Network for 3D Vehicle Detection

3D Vehicle Object Tracking Algorithm Based on Bounding Box Similarity Measurement

Effectively extracting features from a discrete point cloud is necessary for three-dimensional (3D) vehicle tracking. However, a point cloud data set is large and sparsely distributed, hindering the success of vehicle-tracking algorithms. To solve this problem, this paper proposes a 3D vehicle object tracking algorithm based on bounding box similarity measurement. The algorithm includes state prediction, temporal association, trajectory management, state update, and other processes. Also incorporated is a vehicle object temporal association method based on a siamese encoder. The bounding box is encoded into a high-dimensional space, the feature distance is calculated as the time series association cost, and triplet loss was introduced to urge the encoder to learn the geometric similarity of the truth matching box. A 3D Kalman filter and greedy matching are used to effectuate 3D vehicle object tracking algorithm. The experimental results using a KITTI multi-object tracking dataset show that the proposed algorithm can achieve good vehicle tracking performance. In the case of frame loss of point cloud in the frequency reduction simulation, compared with the benchmark method AB3DMOT, the average multi-object tracking accuracy ( AMOTA) and the average multi-object tracking accuracy ( AMOTP) of the improved method are increased by 2.62% and 1.41% respectively, and the performance is better in the frame loss scene.

Vehicle Model and Owner Registration, Detection & Car Price Prediction

Abstract: The key algorithm in autonomous driving is vehicle recognition based computer vision, which tries to recognize whether vehicles are being located by digital photos or videos. Identifying "blocks," which represent the position of the vehicle in photos or videos, is the fundamental concept behind vehicle detection. Additionally, this work explores 3D stereo-based vehicle identification methods, which emerged from sophisticated planar vehicle detection perception. In terms of the differences between the feature extraction approach and the perceived results, this research concludes by summarizing the vehicle detection algorithms used in recent years. It offers theories for more thorough investigation of the vehicle detection systems. With the rapid advancement of machine learning techniques, this tool is now even more essential for object detection. Compared to manually developed features, machine learning-based picture features are more representative. This review paper focuses on object recognition methods based on machine learning and other deep convolutional neural networks however the traditional object detection techniques will also be briefly reviewed.

Development and research of deep neural network fusion computer vision technology

Abstract Deep learning (DL) has revolutionized advanced digital picture processing, enabling significant advancements in computer vision (CV). However, it is important to note that older CV techniques, developed prior to the emergence of DL, still hold value and relevance. Particularly in the realm of more complex, three-dimensional (3D) data such as video and 3D models, CV and multimedia retrieval remain at the forefront of technological advancements. We provide critical insights into the progress made in developing higher-dimensional qualities through the application of DL, and also discuss the advantages and strategies employed in DL. With the widespread use of 3D sensor data and 3D modeling, the analysis and representation of the world in three dimensions have become commonplace. This progress has been facilitated by the development of additional sensors, driven by advancements in areas such as 3D gaming and self-driving vehicles. These advancements have enabled researchers to create feature description models that surpass traditional two-dimensional approaches. This study reveals the current state of advanced digital picture processing, highlighting the role of DL in pushing the boundaries of CV and multimedia retrieval in handling complex, 3D data.

Dynamic response analysis of vehicle–pavement-coupled system based on RIOHTrack full-size accelerated loading test

The interaction between vehicles and asphalt pavements can cause different degrees of damage to the pavement structure. Due to this interaction, deformation of pavement is commonly observed, potentially affecting ride quality and driving safety. Investigating the responses of the vehicle–pavement interaction to different pavement structures is therefore a curial task. To this end, we proposed a vehicle–pavement-coupled dynamic model, based on the accelerated loading test of the Research Institute of Highway MOT Track (RIOHTrack). This model is derived by coupling a heavy-duty truck vibration model and the well-known huge multi-layer rectangular plate model. The heavy-duty truck vibration model which describes a 3D vehicle–pavement dynamics system is constructed according to test conditions. We formulate the coupling based on the variable-order fractional Kelvin foundation and the random dynamic load. In particular, to obtain the random dynamic load, we first analyze the time domain equation of road roughness from the field data set of the RIOHTrack test track and then use the Galerkin method for numerical calculation. Using the proposed model, we simulated 19 types of pavement structures of the RIOHTrack test track, which characterize the primary features of the empirical data set. Furthermore, we discuss the bifurcation consensus of the dynamic response of the front and rear wheels under different loads and analyzes comprehensively the dynamic responses of different pavement structures. The results show that the vehicle–pavement-coupled dynamic response model proposed in this paper has better applicability to simulate deformation for different asphalt pavement structures and better reflects the history-dependent process and memory effect of pavement deformation.

Methodology of fabricating 3D printing modified plastic single-layer panels with UAV positioning technology in the era of mass customization

Mass customization of prefabricated architecture is becoming increasingly crucial for developing the architectural industry. Advanced technologies such as 3D printing and unmanned aerial vehicles (UAV) has brought opportunities and challenges for traditional fabrication and construction methodology. Based on these emerging digital design tools and intelligent construction methods, this paper presents a novel methodology for fabricating single-layer 3D printing panels using UAV positioning technology, which has the potential to revolutionize the construction process and enhance the overall efficiency. This paper first provides a comprehensive review of the existing technologies in 3D printing and UAV positioning, highlighting their benefits and limitations in the context of construction applications. Next, a step-by-step process for fabricating single-layer 3D printing panels is introduced, detailing the optimal design parameters, material selection, and printing techniques. The utilization of UAV for precise positioning and alignment of the panels is then discussed, including the development of an on-site installation for accurate control. To validate the proposed method, a construction practice of the Chengdu Agricultural Expo Centre is produced o demonstrate the promising manufacturing and installation of single-layer 3D printed panels using UAV positioning technology. The results indicate that this method significantly reduces construction time, material waste, and labour costs, while also demonstrating significant customization and design flexibility.

O13 JAK inhibitors to restore skin barrier function in ARCI

Abstract The autosomal recessive congenital ichthyoses (ARCI) are a group of chronic conditions where there is unmet therapeutic need. Harlequin ichthyosis (HI) is the most severe form of ARCI, with aberrant differentiation of the epidermis leading to a severe barrier defect. It is caused by mutations in ABCA12, a lipid transporter involved in the transport of glucosylceramides from the lamellar body to the lipid lamellae. Lamellar ichthyosis is caused by mutations in TGM1. Several Janus kinase (JAK) inhibitors (JAKi) inhibiting the JAK–STAT pathway are now licensed for atopic dermatitis. Previous work from our group in an in vitro HI model showed that tofacitinib had a restorative effect on the lipid barrier. We hypothesized that JAKi A, B and C could restore the epidermal barrier in both HI three-dimensional (3D) organotypic models (OTs) made using ABCA12 CrispR-Cas9 knockout and wild-type keratinocytes and ARCI 3D OTs generated with short hairpin RNA knock down of TGM1. The JAKi were tested at three concentrations in the 3D models (20, 50 and 500 nmol L–1 for JAKiA; 10, 50 and 250 nmol L–1 for JAKiB; 2, 25 and 125 nmol L–1 for JAKiC) and vehicle only was used as the control. The OTs were stained with haematoxylin and eosin, Nile red and differentiation markers such as K10, K14, transglutaminase 1 (TGM1) and involucrin. All three JAKi had a positive effect on the epidermal barrier. The epidermal structure looked more organized and uniform, with an increase in neutral and polar lipids in the upper epidermis. The differentiation markers showed normalization of the expression of involucrin, K10 and TGM1 at the same drug concentrations. These results suggest that JAKi have a generalized positive effect on the skin barrier and will be further investigated in an in vivo mouse model of HI.

3D Modeling & Analysis Techniques for the Apollo Temple in Delphi

This paper demonstrates the application and usage of advanced 3D modeling techniques on monuments through the Apollo Temple in Delphi case study. Firstly, it combines 3D scanning and unmanned aerial vehicle (UAV) photogrammetry to produce an accurate 3D model of the monument, and afterward, it performs finite element modeling (FEM) analysis for both static and dynamic cases. Collapsing scenarios in the case of earthquakes are produced, predicting which parts would first collapse and under which regime the collapsing mechanism would be activated. From the results disclosed herein, the frequency profile of the seismic activity that could lead to resonance with the structure’s dynamic characteristics, and therefore to excessive damages or collapse, was identified. Static structural analysis pinpoints that maximum stress exhibited on the columns’ base never exceeds 0.1338 MPa. Among others, the main novelty of this paper is that it consists of an integrated and multidisciplinary paradigm that advances the available historical knowledge for a quite heavily investigated site.

Bridge Frequency Scanning Using the Contact-Point Response of an Instrumented 3D Vehicle: Theory and Numerical Simulation

Scanning the bridge’s frequencies from the passing vehicle’s vibration data has been frequently investigated recently. However, in previous studies, vehicles were typically simplified to quarter- or half-car models, and apparent disparity could be observed between the models and real vehicles. To make the vehicle model more practical, in this study, a 3D vehicle model is built to extract the bridge’s frequencies from vehicle vibrations. For the first time, equations for calculating the contact-point (CP) response of the 3D vehicle model are derived with tire damping. Furthermore, residual CP responses between front and rear wheels are utilized to eliminate the inverse effects of road roughness, making the bridge frequencies outstanding in the frequency domain. The robustness of the proposed method is tested under different influence factors, and two possible measurement errors are as follows: the sensor position and axle distance when applying the proposed method in engineering. Results show that the proposed method performs stably under the influence of different road roughness classes and tire damping. Bridge frequencies can be identified when the vehicle is travelling at a highway speed (108 km/h in this study). Environmental noises can submerge the bridge’s high-order frequencies but have little influence on the low-frequency range. High bridge damping will restrain the transmission of bridge vibration to the vehicle, making high-order bridge frequencies less visible. In addition, the errors introduced by a vehicle body sensor position can be eliminated when calculating the CP responses for tires, thus will not influence bridge frequency identification. To avoid possible errors induced by manual measurement of the axle distance, a novel cross-correlation function-based method is employed, which is verified effective and practical for calculating residual CP responses.

A comparative study of vehicle-bridge interaction dynamics with 2D and 3D vehicle models

The dynamics of vehicle-bridge interaction (VBI) involve the adoption of two-dimensional (2D) and three-dimensional (3D) vehicle models in simulations. It is worth noting that the 2D vehicle model lacks several crucial degrees of freedom and a complete road roughness compared to the 3D vehicle model. Despite the common usage of the 2D vehicle model in VBI simulations, the difference between the two models has not received sufficient attention in previous studies. Therefore, this study aims to quantitatively assess the differences in bridge response between the 2D and 3D vehicle models. To achieve this goal, the study employs a VBI modeling method based entirely on general finite element software first. Then, a series of comparisons between the 2D and 3D vehicle models are implemented through a steel composite girder bridge and a cable-stayed bridge. The results indicate that road roughness significantly contributes to the observed differences. Specifically, the 2D vehicle model generates larger bridge responses than the 3D vehicle model, primarily due to the absence of rolling and vibration cancellation effects. Furthermore, the differences between the two models increase as the road surface roughness worsens, with a maximum difference of 25.07 % in the cable-stayed bridge’s mid-span acceleration. Notably, the discrepancies in bridge acceleration are more significant than those in displacement or cable stress. Adopting the 2D vehicle model in VBI simulations may lead to more conservative engineering recommendations, resulting in higher economic costs and greater consumption of building materials. Thus, this study emphasizes the importance of considering the 3D vehicle model in VBI simulations, particularly for bridges with poor road conditions.

A Novel Method to Generate Auto-Labeled Datasets for 3D Vehicle Identification Using a New Contrast Model

Auto-labeling is one of the main challenges in 3D vehicle detection. Auto-labeled datasets can be used to identify objects in LiDAR data, which is a challenging task due to the large size of the dataset. In this work, we propose a novel methodology to generate new 3D based auto-labeling datasets with a different point of view setup than the one used in most recognized datasets (KITTI, WAYMO, etc.). The performance of the methodology has been further demonstrated with the development of our own dataset with the auto-generated labels and tested under boundary conditions on a bridge in a fixed position. The proposed methodology is based on the YOLO model trained with the KITTI dataset. From a camera-LiDAR sensor fusion, it is intended to auto-label new datasets while maintaining the consistency of the ground truth. The performance of the model, with respect to the manually labeled KITTI images, achieves an F-Score of 0.957, 0.927 and 0.740 in the easy, moderate and hard images of the dataset. The main contribution of this work is a novel methodology to auto-label autonomous driving datasets using YOLO as the main labeling system. The proposed methodology is tested under boundary conditions and the results show that this approximation can be easily adapted to a wide variety of problems when labeled datasets are not available.

Barometer assisted smartphone localization for vehicle navigation in multilayer road networks

Multilayer transport structures, such as elevated roads and tunnels, are increasingly used to solve the growing traffic congestion problems in dense urban areas. It presents a challenge for navigation systems to determine which layer the vehicle is running on when elevated roads or tunnels are parallel to and close to surface roads. In this paper, we propose a smartphone-based positioning method to provide accurate position and road layer for vehicle navigation in multilayer road networks. Systemic errors of barometers are calibrated with GNSS measurements, local Digital Surface Model (DSM), and Mean Sea Level pressure. Then, barometer output absolute altitudes are integrated with GNSS and IMU to achieve 3D vehicle positioning. Using absolute altitudes, DSM, and GNSS information, the road layer can be identified without initial layer information in multilayer road networks. Experiment results demonstrate that the proposed method can provide accurate location and road layer in multilayer road networks.

Mechanistic Excess Fuel Consumption of a 3D Passenger Vehicle on Rough Pavements

A passenger vehicle’s fuel-powered engine compensates for its dissipated energy, which is caused by road roughness. This paper presents an integrated vehicle–pavement approach and investigates the excess fuel consumption of a seven-degree-of-freedom (DOF) full-car model on rough pavements. In the approach, a nonstationary Laplace process is used to artificially generate road roughness profiles with consideration of local roughness variance and parallel roughness correlation. The proposed mechanistic approach quantifies the impact of road roughness and vehicle dynamic characteristics on excess fuel consumption. The study concluded that overlooking local roughness variance, an indicator of road roughness nonstationarity, may underestimate excess fuel consumption by 30%. Compared with a two-dimensional (2D) half-car model, a three-dimensional (3D) full-car model reduces computation error of excess fuel consumption by approximately 11%. For practical implementation, the extended roughness-speed-impact (ERSI) model (R2=98%) is developed to estimate roughness-induced energy dissipation of five two-axle vehicles.

Mayonnaise-like high internal phase Pickering emulsions stabilized by co-assembled phosphorylated perilla protein isolate and chitosan for extrusion 3D printing application

In this study, the feasibility of the mayonnaise-like high internal phase Pickering emulsions (HIPPEs) prepared by phosphorylated perilla protein isolate-chitosan composite nanoparticles (LZPI-CS CNPs) as three-dimensional (3D) printing inks and lipid-soluble β-carotene delivery vehicles was explored. The spherical LZPI-CS CNPs (232–336 nm in particle size) were formed by hydrogen bonds and electrostatic interaction, which exhibited improved wettability and superior capacity to reduce interfacial tension and enhance the viscoelasticity of the interfacial film even at low chitosan (CS) concentration. The rigid and dense interfacial architecture formed by LZPI-CS CNPs provided steric barrier and strong electrostatic repulsion, preventing droplets from coalescence during long-term storage, centrifugation and heat treatment. The incorporation of CS also obviously reduced the adsorption percentage of phosphorylated perilla protein isolate (LZPI) at the oil-water interface (from 87.27% to 83.15%), and caused the changes of its secondary structure and tertiary conformation. More interestingly, increasing the CS concentration (0.1–0.6 wt%) in LZPI-CS CNPs led to a progressive strengthening of viscosity, viscoelasticity and thixotropy-recovery capacity of HIPPEs, allowing their controllable injectability and printability during 3D printing. The HIPPEs prepared by LZPI-CS CNPs containing 0.6 wt% CS exhibited the most favorable shape fidelity, printing resolution and clear outline. Additionally, the stability of β-carotene against different environmental stresses was also significantly enhanced after encapsulation with HIPPEs. These findings demonstrated that the complexation of LZPI and CS could provide an effective strategy to construct mayonnaise-like HIPPEs as 3D printing inks and delivery vehicles of hydrophobic bioactive components. • The interfacial properties of LZPI could be adjusted by the concentration of CS. • HIPPEs stabilized by LZPI-CS CNPs exhibited superior stability. • The addition of CS improved the viscosity and stiffness of HIPPEs. • 3D printing cylinder based on HIPPEs containing 0.6 wt% CS showed high resolution and clear outline. • The stability of β-carotene was enhanced after HIPPEs encapsulation.