Search Results

This paper deals with an integrated motor-transmission (IMT) speed tracking control of the connected vehicle when there are controller area network (CAN)-induced delays and denial of service (DOS)-induced delays. A connected vehicle equipped with an IMT system may be attacked through the external network. Therefore, there are two delays on the CAN of the connected vehicle, which are CAN-induced and cyber-attack delays. A DOS attack generates huge delays in CAN and even makes the control system invalid. To address this problem, a robust dynamic output-feedback controller of the IMT speed tracking system considering event-triggered detectors resisting CAN-induced delays and DOS-induced delays is designed. The event-triggered detector is used to reduce the CAN-induced network congestion with appropriate event trigger conditions on the controller input and output channels. CAN-induced delays and DOS-induced delays are modeled by polytopic inclusions using the Taylor series expansion.

The development of intelligent and networked vehicles has enhanced the demand for precise road information perception. Detailed and accurate road surface information is essential to intelligent driving decisions and annotation of road surface semantics in high-precision maps. As one of the key parameters of road information, road roughness significantly impacts vehicle driving safety and comfort for passengers. To reach a rapid and accurate estimation of road roughness, in this study, we develop a neural network model based on vehicle response data by optimizing a long-short term memory (LSTM) network through the particle swarm algorithm (PSO), which fits non-linear systems and predicts the output of time series data such as road roughness precisely. We establish a feature dataset based on the vehicle response time domain data that can be easily obtained, such as the vehicle wheel center acceleration and pitch rate.

An improved anti-icing design with film heating ejection slot and cover for the inlet part of aero-engine was brought out, which combines the interior jet impingement with the exterior hot air film heating and shows promising application for those parts manufactured with composite materials. A hybrid method based on the combination of the Back Propagation Neural Network (BPNN) and Genetic Algorithm (GA) is developed to optimize the anti-icing design for a typical aero-engine inlet vane in two dimensions. The optimization aims to maximize the heating performance of the hot air film, which is assessed by the heating effectiveness. The film-heating ejection angle and the cover opening angle are the two geometric variables to be optimized. Numerical model was established and validated to generate training and testing samples for BPNN, which was used to predict the objective function and find the optimal design variables in conjunction with the GA.

Traffic light detection has great significant for unmanned vehicle and driver assistance system. Meanwhile many detection algorithms have been proposed in recent years. However, traffic light detection still cannot achieve a desirable result under complicated illumination, bad weather condition and complex road environment. Besides, it is difficult to detect multi-scale traffic lights by embedded devices simultaneously, especially the tiny ones. To solve these problems, this paper presents a robust vision-based method to detect traffic light, the method contains main two stages: the region proposal stage and the traffic light recognition stage. On region proposal stage, we utilize lane detection to remove partial background from the original image. Then, we apply adaptive canny edge detection to highlight region proposal in Cr color channel, where red or green color proposals can be separated easily. Finally, extract the enlarged traffic light RoI (Region of Interest) to classify.

Forward obstacles detection is one of the key tasks in the perception system of autonomous vehicles. The perception solution differs from the sensors and the detection algorithm, and the vision-based approaches are always popular. In this paper, an embedding fast obstacles detection algorithm is proposed to efficiently detect forward diverse obstacles from the image stream captured by the monocular camera. Specifically, our algorithm contains three components. The first component is an object detection method using convolution neural networks (CNN) for single image. We design a detection network based on shallow residual network, and an adaptive object aspect ratio setting method for training dataset is proposed to improve the accuracy of detection. The second component is a multiple object tracking method based on correlation filter for the adjacent images.

In recent years, due to its strong plasticity and excellent future potential, the development of automated vehicles in the mining environment is extraordinarily rapid. The application of lidar in automated vehicles has also become more and more popular, and algorithms for point cloud object detection have emerged endlessly. However, due to rough road and dusk-to-dawn operations in the mining scenario and the large size of the truck, traditional detection algorithms fail to meet the requirements of real-time updates. Due to the high precision and efficiency of the deep learning network, its application could break through the limitations of traditional algorithms. In this paper, the algorithm called PointPillars is adapted for object detection in the mining scenario. After converting the point cloud to a sparse pseudo-image and extracting features by a two-dimensional convolutional neural network (2D CNN), the time consumption of the entire algorithm becomes much less.

The main objective of this work is to investigate, by means of numerical simulations, the performance of the engine nacelle ventilation cooling system of a helicopter under hover and forward flight conditions, and to propose a simplified method of evaluating the performance based on rotor downwash flow by taking the synthetical effect of engine nacelle, exhaust ejector and external flow of a helicopter into account. For the engine nacelle of a helicopter, an integrated model of the nacelle and exhaust ejector was set up including the domain of external flow. The unstructured grid and finite volume method were applied for domains and control equations discreteness, and the standard k-ε model was applied for solving turbulent control equations. Using the business CFD software, the flow field and the temperature field in the nacelle were calculated for single inlet scheme and double inlets scheme, total up to 9 schemes. The performance of the exhaust ejector was computed.

Recognizing road conditions using onboard sensors is significant for the performance of intelligent vehicles, and the road profile is a widely accepted representation both in the temporal and frequency domains, greatly influencing driving quality. In this paper, a recurrent neural network embedded with attention mechanisms is proposed to reconstruct the road profile sequence. Firstly, the road and vehicle sensor signals are obtained in a simulated environment by modeling the road, tire, and vehicle dynamic system. After that, the models under different working conditions are trained and tested using the collected data, and the attention weights of the trained model are then visualized to optimize the input channels. Finally, field experiments on the real vehicle are conducted to collect real road profile data, combined with vehicle system simulation, to verify the performance of the proposed method.