Refine Your Search

Topic

Author

Affiliation

Search Results

Technical Paper

Neural-Network-Based Suspension Kinematics and Compliance Characteristics and Its Implementation in Full Vehicle Dynamics Model

2022-03-29
2022-01-0287
Suspension kinematics and compliance strongly influence the handling performance of the vehicle. The kinematics and compliance characteristics are determined by the suspension geometry and stiffness of suspension linkage and elastic components. However, it is usually inefficient to model all the joints, bushings, and linkage deformation in a full vehicle model. By transforming the complex modeling problem into a data-driven problem tends to be a good solution. In this research, the neural-network-based suspension kinematics and compliance model is built and implemented into a 17 DOF full vehicle model, which is a hybrid model with state variables expressed in the global coordinate system and vehicle coordinate system. The original kinematics and compliance characteristics are derived from multibody dynamics simulation of the suspension system level.
Technical Paper

Virtual Evaluation of Deep Learning Techniques for Vision-Based Trajectory Tracking

2022-03-29
2022-01-0369
Ground vehicle systems are still largely reliant on traditional control technique deployments. Deep-learning based control-system deployments, are emerging as a viable substitute to more traditional control-system. However, reliability and robustness of deep-learning based controllers in actual deployments and their subsequent verification and validation remains a challenge. This is exacerbated by the need to factor in the uncertainty of the environment as well as the increased number of parameters. Existing literature comparisons of deep-learning vs traditional controllers do not offer structured approaches to performance evaluation and improvement. It is also crucial to: (i) develop a standardized controlled test-environment within which various controllers are evaluated against a common metric; (ii) identify a reference high-fidelity controller (traditional) that can serve as a benchmark.
Technical Paper

Application of a Digital Twin Virtual Engineering Tool for Ground Vehicle Maintenance Forecasting

2022-03-29
2022-01-0364
The integration of sensors, actuators, and real-time control in transportation systems enables intelligent system operation to minimize energy consumption and maximize occupant safety and vehicle reliability. The operating cycle of military ground vehicles can be on- and off-road in harsh weather and adversarial environments, which demands continuous subsystem functionality to fulfill missions. Onboard diagnostic systems can alert the operator of a degraded operation once established fault thresholds are exceeded. An opportunity exists to estimate vehicle maintenance needs using model- based predicted trends and eventually compiled information from fleet operating databases. A digital twin, created to virtually describe the dynamic behavior of a physical system using computer-mathematical models, can estimate the system behavior based on current and future operating scenarios while accounting for past effects.
Technical Paper

Road Adhesion Coefficient Identification Method Based on Vehicle Dynamics Model and Multi-algorithm Fusion

2022-03-29
2022-01-0908
As an important input parameter of intelligent vehicle active safety technology, road adhesion coefficient is of great significance in autonomous collision avoidance, emergency braking and collision avoidance, and variable adhesion road motion control. Traditional recognition methods based on vehicle dynamics require large data volume and low solution accuracy. This paper proposes an adhesion coefficient recognition method based on Elman neural network and Kalman filter. By establishing a seven-degree-of-freedom vehicle dynamics model, dynamic parameters such as yaw angular velocity, longitudinal velocity, lateral velocity, and angular velocity of each wheel, which are easy to measure and strongly related to the road adhesion coefficient, are analyzed as the input of the neural network model.
Technical Paper

Research on Brake Comfort Based on Brake-by-Wire System Control

2022-03-29
2022-01-0912
The vehicle will produce certain shock and vibration during the braking process, which will affect the driving experience of the driver. Aiming at the problems of pitch vibration, longitudinal vibration and shock during the braking process, this paper proposes a planning and following control method for target longitudinal acceleration in post-braking phase, and designs control trigger strategies. Target longitudinal acceleration planning takes minimizing longitudinal shock as the design goal. The following control takes the brake pressure as the control object, and adopts the "feedforward +PID feedback" method to follow the target longitudinal acceleration. Besides, considering the safety of braking process, the trigger condition of control is designed which utilizes BP neural network method to judge whether the control has to be triggered. Based on Simulink software, the simulation model of straight-line braking is established.
Technical Paper

Adjoint-Based Model Tuning and Machine Learning Strategy for Turbulence Model Improvement

2022-03-29
2022-01-0899
As turbulence modeling has become an indispensable approach to perform flow simulation in a wide range of industrial applications, how to improve the prediction accuracy has gained increasing attention during the past years. Of all the turbulence models, RANS is the most common choice for many OEMs due to its short turn-around time and strong robustness, however, the default setting of RANS is usually benchmarked through classical and well-studied engineering examples, not always suitable for resolving complex flows in specific applications. Many previous researches have suggested a small tuning in turbulence model coefficients could achieve higher accuracy on a variety of flow scenarios. Instead of adjusting parameters by trial and error from experience, this paper introduced a new data-driven approach of turbulence model recalibration using adjoint solver, based on Generalized k-ω (GEKO) model, one variant of RANS.
Technical Paper

Development of a Prediction Model for Tire Tread Pattern Noise based on Convolutional Neural Network with RMSProp Algorithm.

2022-03-29
2022-01-0884
It has changed from an internal combustion engine to an electric vehicle(EV) so the exterior noise of the engine and exhaust system is removed and the tire pattern noise becomes the main noise source contributor for EV traveling at speeds above 50 kph and 80 kph. It is very useful to predict tire pattern noise performance in the early tire design stage. An artificial neural network model was recently used for the prediction of tire pattern noise. The ANN used the supervised training method, in which features are extracted by applying Gaussian curve fitting to the tread profile spectra of tire patterns; these are then used as the inputs of the ANN. Nevertheless, this method requires the laser scanning of the patterns of real tires, which are non-existent as they are still in the early design stage.
Technical Paper

A Multi-axle and Multi-Type Truck Classification Model for Dynamic Load Recognition

2022-03-29
2022-01-0137
Overloading of trucks can easily cause damage to roads, bridges and other transportation facilities, and accelerate the fatigue loss of the vehicles themselves, and accidents are prone to occur under overload conditions.In recent years, various countries have formulated a series of management methods and governance measures for truck overloading.However, the detection method for overload behavior is not efficient and accurate enough.At present, the method of dynamic load identification is not perfect.
Technical Paper

Securing camera based active driver monitoring system from video forgery attacks using deep learning

2022-03-29
2022-01-0115
The numerous superiorities of autonomous vehicles in terms of safety, driving experience, and comfort against the traditional driving favor them in the wide adoption across the modern automotive sector. Driver Monitoring System (DMS) is one of the high Automotive Safety Integrity Level (ASIL) specified driver assistance functionalities, which assists the driver continuously as a part of Active Safety system. The fused vision-related functionalities of the camera based active DMS such as Distraction, Drowsiness, and Fatigue detection systems monitors the driver’s Head movements, Eye movements, and Facial emotions respectively to awake the driver with constant alerts under undesired conditions. The presence of the modern day DMS in the In-Vehicle-Infotainment Digital Cockpits exposes the critical DMS into a wide range of cyber-attacks either locally or remotely, which in turn causes the malfunctioning of the active safety driver assistance system.
Technical Paper

Optimization of Computer Vision Software Models for Deployment to Performance Constrained Embedded Processors

2022-03-29
2022-01-0160
Computer vision (CV), a form of artificial intelligence (AI), is a foundational technology within the automotive industry for an increasing number of applications including active safety, motion control, and driver distraction monitoring. State-of-the-art CV models often rely on the use of Deep Neural Networks (DNNs) to achieve high levels of accuracy. While necessary for their accuracy, DNNs are computationally complex. For example, when compared to other AI model architectures, they have a large memory footprint and must perform a high number of operations to create an output or prediction. To meet performance goals in the face of such constraints, high performance processors such as Graphics Processing Units (GPUs) are typically required to run CV models on-board automobiles, creating a major hurdle to the deployment of CV applications.
Technical Paper

Research on Autonomous Driving Decision Based on Improved Deep Deterministic Policy Algorithm

2022-03-29
2022-01-0161
Autonomous driving technology, as the product of the fifth stage of the information technology revolution, is of great significance for improving urban traffic and environmentally friendly sustainable development. Autonomous driving can be divided into three main modules. The input of the decision module is the perception information from the perception module and the output of the control strategy to the control module. The deep reinforcement learning method proposes an end-to-end decision-making system design scheme. This paper adopts the Deep Deterministic Policy Gradient Algorithm (DDPG) that incorporates the Priority Experience Playback (PER) method. The framework of the algorithm is based on the actor-critic network structure model. The model takes the continuously acquired perception information as input and the continuous control of the vehicle as output.
Technical Paper

A prediction model of RON loss based on neural network

2022-03-29
2022-01-0162
The RON(Research Octane Number) is the most important indicator of motor petrol, and the petrol refining process is one of the important links in petrol production. However, RON is often lost during petrol refining and RON Loss means the value of RON lost during petrol refining. The prediction of the RON loss of petrol during the refining process is helpful to the improvement of petrol refining process and the processing of petrol. The traditional RON prediction method relied on physical and chemical properties, and did not fully consider the high nonlinearity and strong coupling relationship of the petrol refining process. There is a lack of data-driven RON loss models. This paper studies the construction of the RON loss model in the petrol refining process.
Technical Paper

End-to-End Synthetic LiDAR Point Cloud Data Generation and Deep Learning Validation

2022-03-29
2022-01-0164
LiDAR sensors are common in automated driving due to their high accuracy. However, LiDAR processing algorithm development suffers from lack of diverse training data, partly due to sensors’ high cost and rapid development cycles. Public datasets (e.g. KITTI) offer poor coverage of edge cases, whereas these samples are essential for safer self-driving. We address the unmet need for abundant, high-quality LiDAR data with the development of a synthetic LiDAR point cloud generation tool and validate this tool’s performance using the KITTI-trained PIXOR object detection model. The tool uses a single camera raycasting process and filtering techniques to generate segmented and annotated class-specific datasets.
Technical Paper

Research on neural network-based prediction model for CNG engine intake charge

2022-03-29
2022-01-0166
Based on the sample data obtained from the bench test of a four-cylinder naturally aspirated CNG engine, three different neural networks, BP, SVM and GRNN, were used to develop the intake charge prediction model for the intake system of this engine, in which engine speed, intake manifold pressure and intake temperature, VVT angle and gas injection time were taken as input parameters and intake charge was used as output parameter. The comparative analysis of the experimental data and model prediction data showed that the mean absolute error (MAE) of BP model, SVM model, and GRNN model were 2.69, 5.13, and 8.11, and the root mean square error (MSE) were 3.53, 7.17, and 9.29, respectively. BP neural network has smaller prediction error and higher accuracy than SVM and GRNN neural network, which is more suitable for the prediction of the intake charge of this type of four-cylinder naturally aspirated CNG engine.
Technical Paper

An Efficient Machine Learning Algorithm for Valve Fault Detection

2022-03-29
2022-01-0163
Multi-level Miller-cycle Dynamic Skip Fire (mDSF) is a combustion engine technology that improves fuel efficiency by deciding on each cylinder-event whether to skip (deactivate) the cylinder, fire with low (Miller) charge, or fire with a high (Power) charge. In an engine with two intake and two exhaust valves, skipping the cylinder is accomplished by deactivating all valves, while firing with a reduced charge is accomplished by deactivating one of the intake valves. This new ability to modulate the charge level introduces new failure modes. The first is a failure to reactivate the single intake valve, which results in a desired high-fire having the air intake of a low-fire; the second is a failure to deactivate the single intake valve, which results in a low-fire having the air intake of a high-fire.
Technical Paper

Research on vehicle following control based on deep reinforcement learning in intelligent network environment

2022-03-29
2022-01-0165
At present, the deep reinforcement learning algorithm in the field of artificial intelligence is widely used in the engineering field because of its strong self-learning ability, and many scholars have also done research on the application of reinforcement learning algorithm to unmanned control. However, reinforcement learning algorithm needs to be learned through "trial and error" in the environment, which makes the application of this kind of algorithm in the real physical environment more difficult, it is impossible for them to carry out "trial and error" learning on real vehicles. Therefore, the randomness of reinforcement learning algorithm in the early stage of agent learning is a major obstacle to its application in the real physical environment.
Technical Paper

Vehicle Feature Recognition Method Based on Image Semantic Segmentation

2022-03-29
2022-01-0144
In the process of truck overload and over-limit detection, it is necessary to detect the characteristics of the vehicle's size, type, and wheel number. In addition, in some vehicle vision-based load recognition systems, the vehicle load can be calculated by detecting the vibration frequency of specific parts of the vehicle or the change in the length of the suspension during the vehicle's forward process. Therefore, it is essential to quickly and accurately identify vehicle features through the camera. This paper proposes a vehicle feature recognition method based on image semantic segmentation and Python, which can identify the length, height, number of wheels and vibration frequency at specific parts of the vehicle based on the vehicle driving video captured by the roadside camera.
Technical Paper

Road Crossing Assistance Method Using Object Detection Based on Deep Learning

2022-03-29
2022-01-0149
This paper describes a method for assisting pedestrians to cross a road. As motorization develops, pedestrian protection techniques are becoming more and more important. Advanced driving assistance systems (ADAS) are improving rapidly to provide even greater safety. However, since the accident risk of pedestrians remains high, the development of an advanced walking assistance system for pedestrian protection may be an effective means of reducing pedestrian accidents. Crossing a road is one of the highest risk events, and is a complex phenomenon that consists of many dynamically changing elements such as vehicles, traffic signals, bicycles, and the like. A road crossing assistance system requires three items: real-time situational recognition, a robust decision-making function, and reliable information transmission. Edge devices equipped with autonomous systems are one means of achieving these requirements.
Technical Paper

Accelerating In-vehicle Network Intrusion Detection System using Binarized Neural Network

2022-03-29
2022-01-0156
Modern vehicles are utilizing more software modules and interfaces while new risks and attacks are emerging and threatening the security of vehicles. Researchers demonstrate that in-vehicle networks have become a target of vehicle attackers. Controller Area Network (CAN), the de facto standard for in-vehicle networks, has insufficient security features and thus is inherently vulnerable to various attacks. Intrusion detection systems (IDSs) based on advanced deep learning methods, such as Convolutional Neural Network (CNN) and Recurrent Neural Network (RNN), have been proposed to protect CAN bus from attacks. However, those models generally introduce high latency and require considerable computing resources. To accelerate intrusion detection and also reduce both latency and model size, we propose a new IDS system based on Binarized Neural Network (BNN).
Technical Paper

Databased modeling: An AI Toolchain for the development process of combustion engines

2022-03-29
2022-01-0158
Predictive physical modeling is an established method used in the development process for any automotive component or system. While accurate predictions can be issued after tuning model parameters, long computation times can be expected depending on the complexity of the model at hand. As the requirements of the components/systems to be developed continuously increase, new optimization approaches are constantly being applied to solve multidimensional objectives and resulting conflicts optimally. Unfortunately, some of those approaches are deemed not feasible, as the computational times of the required single predictions using conventional simulation models are too high. Therefore, it is proposed to use data-based models such as trained neural networks instead of physical models to address this issue. Previous efforts have failed due to a weak database and the resulting poor predictive ability of data-based models.
X