Search Results

Using highly efficient powertrain is one of the most important and effective approaches to increase the driving distance of electric vehicles (EVs). In this paper, a novel two-speed dual-clutch transmission (DCT) is proposed. The transmission is comprised of two traditional friction clutches and two-stage planetary gear sets. One clutch connects the input sun gear and the other connects the input carrier. The Simulink models including an electric motor and two-speed DCT are established. Gearshift schedule based on fuzzy logic which reflects the driver’s intensions is adopted to improve the dynamic and economic performance of the novel transmission. The simulation model is built using MATLAB/Simulink® to validate the effectiveness of the proposed gearshift schedule compared with the conventional two-parameter gearshift schedule. Simulation results show that both the dynamic and economic performance of the novel DCT for EVs are improved with the proposed fuzzy logic gearshift schedule.

Vehicle lateral states such as lateral distance at a preview point and heading angle are indispensable for lane keeping control systems, and such states are normally estimated by fusing signals from an onboard vision system and inertial sensors. However, the sampling rates and measurement delays are different between the two kinds of sensing devices. Most of the conventional methods simply neglect measurement delay and reduce sampling rate of the estimator to adapt to the slow sensors/devices. However, the estimation accuracy is deteriorated, especially considering the delay of visual signals may not be constant. In case of electric vehicles, the actuators for steering and traction are motors that have high control frequency. Therefore, the frequency of vehicle state feedback may not match the control frequency if the estimator is infrequently updated. In this paper, a multi-rate estimation algorithm based on Kalman filter is proposed to provide lateral states with high frequency.

Precise trajectory prediction of surrounding vehicles is critical for decision-making of autonomous vehicles, and learning-based approaches are well recognized for the robustness. However, state-of-the-art learning-based methods ignore (1) the feasibility of the vehicle’s multimodal state information for prediction and (2) the mutually exclusive relationship between the global traffic scene receptive fields and the local position resolution when modeling vehicles’ interactions, which may influence prediction accuracy. Therefore, we propose a “vehicle descriptor”-based long short-term memory (LSTM) model with the dilated convolutional social pooling (VD+DCS-LSTM) to cope with the above issues.

A novel magnetorheological fluid dual clutch (MRFDC) for electric vehicle transmission is proposed in this article. The structure was based on the MR fluid clutch and traditional dual clutch equipped on internal combustion engine vehicle. Therefore the MRFDC combines the advantages of MR fluid clutch and dual clutch transmission (DCT) to achieve high control accuracy and fast response. The structure of MRFDC was designed by Unigraphics (UG) three-dimensional (3D) modeling software. Then, finite element analysis (FEA) for magnetic field was conducted by ANSYS under different applied currents from 0.1A to 1A with 0.1A space to obtain the relation between the applied current and magnetic field. In this article, Herschel-Bulkley model is used to predict the MR fluid behavior because of the high shear rate of MR fluid.

Visual Place Recognition (VPR) excels at providing a good location prior for autonomous vehicles to initialize the map-based visual SLAM system, especially when the environment changes after a long term. Condition change and viewpoint change, which influences features extracted from images, are two of the major challenges in recognizing a visited place. Existing VPR methods focus on developing the robustness of global feature to address them but ignore the benefits that local feature can auxiliarily offer. Therefore, we introduce a novel hierarchical place recognition method with both global and local features deriving from homologous VLAD to improve the VPR performance. Our model is weak supervised by GPS label and we design a fine-tuning strategy with a coupled triplet loss to make the model more suitable for extracting local features.