16 January 2024 in IEEE Transactions on Transportation Electrification
As the preferred configuration for autonomous vehicles, the steer-by-wire (SbW) system replaces the mechanical linkage with electric signals. However, the disconnection of the steering column deteriorates the origin steering feedback torque (SFT), necessitating the road sense construction. To provide the potential driver convincing steering feel, this paper proposes a road sense construction scheme, including an SFT designer employing the data-driven method and an SFT controller based on the friction compensated explicit model predictive control (FEMPC). Initially, the hand wheel module and road wheel module are introduced to analyze network features and establish controller foundations. Subsequently, the torque designer combining convolution neural network (CNN) and gated recurrent unit (GRU) layers is proposed, which can extract the spatial and temporal information of vehicle dynamic (VD) states respectively. To ensure precision and real-time performance, the torque controller based on the explicit model predictive control (EMPC) is then presented and the Brown friction model is also adopted for feedforward compensation. After training the designer with real vehicle data, the whole road sense construction scheme is verified subjectively and objectively. The simulation and experiment results show the authenticity of the designer on the one hand and the precision of the controller on the other.
15 November 2023 in IEEE Transactions on Industrial Informatics
The steer-by-wire (SbW) system has gained recognition as the future of intelligent vehicles due to its attributes, such as safety, simplification, and flexibility. However, the elimination of mechanical linkage necessitates the provision of artificial steering feedback torque (SFT), which is crucial for the potential driver to manipulate the vehicle. To enhance the steering feel, this article extends the SFT design range and proposes an SFT prediction scheme based on the sequence-to-sequence (S2S) network with the switcher-assisted (SA) training algorithm. The models of the electric power steering (EPS) and SbW are first established to analyze the input features. The S2S network with gated recurrent units (GRU) is then presented, where the encoder scheme incorporates the squeeze-and-excitation (SE) operation to achieve adaptive feature recalibration. Subsequently, the SA algorithm is proposed to alleviate the exposure bias based on the principle of multitask learning (MTL), wherein online training is regarded as the main task and offline training is treated as the auxiliary task. The weighting coefficients between tasks are optimized using an assisted network named switcher, facilitating a gradual transition from MTL to the single main task, thereby avoiding complex tuning processes. The validation results indicate the proposed scheme outperforms existing methods regarding estimation and prediction. The ablation experiments are further conducted to illustrate the effectiveness of SE blocks and the SA algorithm. Finally, the SFT construction simulation, involving target prediction and torque tracking, is conducted, validating that variable-length prediction can adapt to various conditions and improve tracking performance.
17 August 2023 in Proceedings of the Institution of Mechanical Engineers, Part D: Journal of Automobile Engineering
Four-wheel steering (4WS) is an advanced technology for automotive chassis, which is more flexible and redundant than conventional front wheel steering. For a 4WS vehicle, the accuracy and stability of steering control must be guaranteed, so the main difficulty is the model nonlinearity during driving. To deal with the problems, a novel composite model free observer (CMFO) is designed. The nominal parameters are used to model the 4WS vehicles, and the data-driven method is used to improve observation accuracy. Then a CMFO based controller is designed to control the lateral dynamics of 4WS vehicles. The stability of the proposed control scheme is proved by theoretical derivation, and experiments are carried out to compare the performance between the CMFO based controller, model free adaptive controller (MFAC) and linear quadratic regulator (LQR). The mean absolute error (MAE) values of CMFO are reduced by 75.56% and 44.49% compared with those of LQR and MFAC, respectively.
15 May 2025 in Proceedings of the Institution of Mechanical Engineers, Part D: Journal of Automobile Engineering
The escalating demands for software-defined vehicles have prompted advancements in automated driving, facilitating a shift toward a novel chassis domain electronics architecture equipped with control-by-wire actuators. How to bolster driving dynamics and stability performance via chassis domain control represents a significant and challenging endeavor. This article presents a novel dual-layer adaptive active front steering and electronic stability control (AFS-ESC) coordinated control architecture based on the robust nonsingular fast terminal sliding mode control (NFTSMC) method, aimed at stabilizing the vehicle and enhancing the maneuverability. Initially, the robust NFTSMC theory is rigorously derived and analyzed for uncertain nonlinear multiple input multiple output systems. It is then applied to develop yaw rate-based maneuverability and side-slip angle-based stability decision-making using vehicle and tire models with unknown nonlinearities, uncertainties, and disturbances. Additionally, an adaptive gain based on a stability index is introduced to coordinate the decision-making outcomes. Subsequently, the requisite total yaw-moment is allocated to the front wheel steering angle and wheel cylinder pressures through AFS and ESC, respectively, based on an adaptive semiempirical tire nonlinearity index. The efficacy of the proposed architecture is validated via 0.7 Hz sine with dwell tests with high, medium, and low tire-road adhesion on the hardware-in-the-loop platform, demonstrating its superior performance in stabilizing the vehicle and enhancing maneuverability compared to the robust control-based and NFTSMC-based AFS-ESC coordination, ESC, AFS, and uncontrolled cases.