This study investigates a simulation-based estimation–control chain that integrates an interactive multiple model adaptive unscented Kalman filter (IMM-AUKF) with a multiple-input multiple-output model predictive controller (MIMO-MPC). A seven-degree-of-freedom vehicle model and a Pacejka tire model are used to represent nonlinear vehicle dynamics. The controller updates model and regression region steering constraints from the estimated adhesion state and jointly allocates front/rear steering and longitudinal force commands. The revised experiments use a 0.5 ms plant integration step and a consistent 20 ms estimator/controller update. Under high-adhesion double lane change (DLC), the proposed chain lowers speed RMSE from 0.7950 to 0.1833 m/s and mean adhesion estimation RMSE from 0.1567 to 0.0812. Under variable-adhesion single lane change (SLC), lateral RMSE decreases from 0.1816 to 0.1649 m, speed RMSE from 0.7983 to 0.2505 m/s, and mean adhesion estimation RMSE from 0.1622 to 0.0949. Heading error is not uniformly improved and is reported as a design trade-off. These results provide reproducible simulation evidence, while hardware-in-the-loop and real-vehicle validation remain future work.
Lane-Keeping Assist Systems (LKAS) are widely used in modern passenger vehicles to support lateral vehicle control and reduce the risk of unintended lane departure. Although LKAS performance is commonly evaluated in relation to perception, control, and sensor fusion, the observable vehicle response may also depend on the mechanical condition of the chassis. This study presents a qualitative, measurement-based evaluation of the influence of intentionally introduced front-wheel toe misalignment on the observable response of a production LKAS under controlled proving-ground conditions. Experimental tests were conducted on the highway module of the ZalaZONE proving ground using a Lexus RX 450h equipped with a factory-installed LKAS function. Three front-wheel toe configurations were investigated: factory-specified alignment, single-wheel toe misalignment, and severe toe misalignment affecting both front wheels. Measurements were performed at 70, 90, and 110 km/h on straight and curved road sections. Vehicle speed, steering angle, lateral acceleration, and GNSS-based position data were recorded using CAN- and GNSS/IMU-based data acquisition. The qualitative comparison of the measured signal profiles indicated that the misaligned configurations were associated with a shifted steering-angle operating range and less uniform steering and lateral-acceleration responses. The most pronounced visible differences occurred under the severe toe-misalignment condition, particularly at higher speeds and in the curved section. As the analysis did not include quantitative effect measures or statistical comparisons, these observations are interpreted as exploratory tendencies rather than statistically validated changes in LKAS performance. The findings suggest that front-wheel toe condition should be considered in the measurement-based assessment, maintenance, and calibration of ADAS-equipped vehicles.
This paper presents an adaptive event-triggered nonsingular fast terminal sliding-mode control (AETSMC) framework for quadrotor unmanned aerial vehicles subject to aerodynamic disturbances, actuator loss of effectiveness (LOE) of up to 60%, and intermittent denial-of-service (DoS) attacks. First, a nonsingular fast terminal sliding-mode (NFTSM) surface is constructed using fractional powers of the tracking error rather than fractional-order derivatives. This design ensures finite-time convergence while avoiding the singularity associated with conventional terminal sliding-mode schemes. Second, a smooth positive-semidefinite barrier function (Smooth-PSBF) is incorporated into the adaptive gain law. The resulting law provides only the compensation required to maintain the prescribed bound, thereby limiting gain overestimation and chattering. Third, a dual-mode event-triggering mechanism combines an exponentially decaying threshold with a zero-order hold. A positive lower bound on the inter-event interval is derived from the closed-loop dynamics, which excludes Zeno behaviour. Simulations under matched conditions show that the proposed method reduces the pitch-channel root-mean-square error by 79.4% and the integral squared error by 95.8% relative to the first reproduced baseline. In a separate 15-s communication experiment sampled at 1 kHz, the controller generated 128 transmissions instead of 15,000 periodic updates, corresponding to a 99.15% reduction. These results indicate that the proposed framework can improve fault-tolerant tracking while reducing communication demand under intermittent DoS attacks.
Ride quality—encompassing vehicle comfort, vibration isolation, and noise, vibration, and harshness (NVH)—has become a key competitive differentiator in modern automobiles. This paper presents a comprehensive update to the foundational literature on passenger car ride quality, capturing the rapidly growing literature and new technological paradigms that have emerged over the past decade. Established approaches to human vibration response, vehicle dynamics modelling, and road surface characterisation are examined within the ISO 2631 framework. This review critically surveys advances driven by battery electric vehicle (BEV) powertrains—where the absence of internal combustion engine noise unmasks motor whine, inverter switching noise, and tyre–road excitation, lowering the perceptual ride–NVH boundary from ~25 Hz toward 15–18 Hz—as well as intelligent semi-active and active suspension technologies, deep reinforcement learning for suspension control, machine learning for ride quality prediction, and connected vehicle infrastructure enabling predictive preview control. Key research gaps are identified: the absence of validated ISO 2631 frequency weightings for autonomous vehicle postures, the lack of standardised open benchmark datasets for cross-study comparison, and the unresolved sim-to-real validation gap for data-driven suspension controllers. Ten priority research directions are proposed for the coming decade.
In light of the advancement of vehicle electrification and intelligence, four-wheel independent drive (4WID) electric vehicles (EVs) have garnered significant attention as a promising platform. Integrating advanced torque-vectoring (TV) strategies into 4WID EVs can effectively optimize the synergistic performance between handling stability and energy efficiency of the over-actuated system across various driving conditions. In this paper, a hierarchical Combined Sliding Mode Control–Adaptive Nonlinear Model Predictive Control (cSMC-ANMPC) TV strategy is proposed to enhance the comprehensive performance of 4WID EVs and ensure adaptive control across diverse driving conditions. Firstly, a hierarchical control architecture is developed to decouple the complex multi-objective problem. The upper layer performs robust stability decision-making by observing the vehicle’s state errors. The lower layer determines the optimal torque distribution throughout the powertrain. Secondly, a Combined Sliding Mode Controller (cSMC) is developed for the upper layer to promptly generate a robust stability command. By co-regulating both yaw rate and sideslip angle into a single command, it simplifies the lower layer’s task and enhances overall stability. Thirdly, a Soft Actor-Critic (SAC) intelligent tuner is integrated into the lower-layer NMPC to mitigate the effects of varying conditions on the stability–economy trade-off and strengthen the adaptability of the controller. Finally, co-simulation evaluations on the MATLAB R2023b/CarSim 2020.0platform demonstrate that the proposed cSMC-ANMPC strategy can improve comprehensive performance for the studied 4WID EV. Compared with other baselines, the stability enhancement in extreme maneuvers and the long-term energy-saving capability are remarkable, showcasing its promising performance.
This article presents a feasibility study on replacing the conventional viscoelastic materials used to mount exhaust systems to the chassis with magnetorheological materials. The objective is to reduce vibrations caused by the road irregularities that affect ride comfort through noise and vibration in the vehicle’s passenger compartment. Magnetorheological materials are already used in semi-active suspension systems to provide variable stiffness. A simple dynamic model is developed that incorporates a magnetorheological actuator, providing a simulation framework for various scenarios and operating conditions. The Bouc–Wen semi-empirical model is used to represent the magnetorheological material. Simulations were performed in accordance with ISO 8608, which defines road profiles for the different simulation scenarios. Extensive simulation tests show that the proposed approach is more effective than the viscoelastic materials commonly used in motor vehicle exhaust systems. This work advances the goal of improved noise and vibration control in vehicles and helps lay the groundwork for future research to address these problems.
Efficient discovery of safety-critical scenarios is a central problem in autonomous-driving simulation because safety-critical events are rare and the scenario parameter space is often high-dimensional. Direct random sampling can spend most of the simulation budget on samples that are weakly relevant, physically unreachable, or behaviorally inconsistent. This study proposes a constraint- and risk-driven framework for safety-critical scenario discovery. The framework first constructs a valid scenario space using physical reachability, behavioral consistency, and traffic-feasibility constraints. It then evaluates valid simulated samples with a hierarchical scenario-value model that combines kinematic criticality, longitudinal controllability risk, and scenario diversity. Finally, a risk-feedback probabilistic search updates the sampling distribution using high-value samples. The method is evaluated primarily in a lead-vehicle hard-braking scenario and additionally in a cut-in scenario. In the lead-vehicle hard-braking scenario, the proposed method improves the discovery rate from 0.68% for Random Sampling to 49.62% under a budget of 1000 evaluations. In the cut-in scenario, the discovery rate improves from 5.28% to 67.68% under the same budget. Ablation results show that constraint-aware construction reduces potentially invalid samples, while the full hierarchical scenario-value model improves coverage of discovered safety-critical samples.
Road traffic accidents pose a significant threat to life and property. However, the multitude of contributing factors makes it challenging to pinpoint the most critical ones, thereby hindering effective prevention strategies. Therefore, this study conducts relevant work using the UK Department for Transport’s 2019 road traffic accident datasets. To investigate the distribution characteristics of accidents across various dimensions (person, vehicle, road, environment, and accident configuration), we first preprocessed the data. Missing values were imputed using a chained random forest-based multiple imputation method. To identify key contributing factors, we employed an integrated approach combining Bayesian-optimized random forest, Cramér’s V correlation test, K-modes clustering, and frequency statistics. This framework enabled the exploratory identification of potential high-risk scenarios for both non-operating and passenger vehicles. Subsequently, we applied a constraint-based Apriori algorithm to analyze correlations across these dimensions and temporal factors, revealing significant associations between accident severity and the examined attributes. Finally, a Bayesian-optimized LightGBM model was built to predict accident risk levels. External validation using the 2022 UK dataset, combined with interpretive analysis, confirmed the model’s strong generalization ability.
Motorcycle-dominated mixed traffic challenges vehicle-behavior modeling because motorcycles and cars differ in footprint, maneuverability, and longitudinal speed regulation. This study used roadside-video trajectories from six arterial sites in Hanoi, Vietnam. The trajectory pipeline produced 202,296 valid vehicle records, and the primary analysis included 158,821 motorcycles and 38,175 private cars. A focused manual identity-continuity audit of two 30 s clips yielded overall MOTA values of 0.9938 and 0.9977 and IDF1 values of 0.9931 and 0.9932. Motorcycles traveled faster on average but showed more stable longitudinal speed profiles: motorcycle speed standard deviation was 45% lower than the car value, and motorcycle coefficient of variation was less than half the car value. The directions of the mean-speed, speed-standard-deviation, and coefficient-of-variation contrasts were consistent across all six sites. Diagnostic reruns using lower speed bounds of 0–5 km/h and minimum trajectory lengths of 10–30 frames retained the same motorcycle–car stability direction (rrb=−0.759 to −0.643 for speed standard deviation and −0.828 to −0.728 for coefficient of variation). A three-component Gaussian mixture model was retained as a parsimonious engineering partition. Repeated-subsample stability was excellent for motorcycles (median adjusted Rand index, ARI, 0.970) and moderate for cars (median ARI 0.706), while mean maximum posterior probabilities exceeded 0.93 for both classes. Savitzky–Golay smoothing substantially increased usable acceleration coverage, but the inferred class contrast changed with the smoothing window; acceleration was therefore excluded from the primary clustering. A high-speed/high-variability rule flagged 4.62% of cars and 0.58% of motorcycles. This rule is an operational kinematic-screening heuristic, not a crash-probability model. The results provide class-specific evidence for mixed-traffic simulation calibration and cautious operational screening under the observed Hanoi arterial conditions.
Adaptive restraint systems require specific occupant information, including head position, anthropometry, and safety-relevant posture states. Existing 3D human pose estimation benchmarks mostly report root-relative pose, while automotive in-cabin studies rarely evaluate these outputs across heterogeneous vehicle sensor sets. We present ICF-Fusion, a five-modality transformer fusion architecture, and evaluate it under leave-one-subject-out (LOSO) validation on the ICF-Body dataset, which includes synchronized near-infrared (NIR) camera, 60 GHz millimeter-wave (mmWave) radar, belt webbing extraction sensor (WES), seat configuration sensor (SCS), and ultra-wideband (UWB) recordings. The model localizes the head with a Mean Root Position Error (MRPE) of 6.10 cm and regresses anthropometry to mean absolute errors (MAE) of 5.36 cm for height, 3.50 cm for torso length, 1.78 cm for shoulder width, and 8.61 kg for weight. Feet-on-dashboard is detected on 9 of 10 evaluable folds without meaningful MRPE degradation. The full sensor fusion outperformed every single modality on all three tasks, but NIR alone nearly matched it for head localization and feet-on-dashboard detection. The fusion advantage was substantial only for the anthropometry estimation task.
This paper presents the research, development, and functional validation of an original rear suspension system designed for hybrid reverse trike vehicles (two guided wheels on the front axle and a twin-tire-driven assembly at the rear). Conventional configurations featuring a single rear wheel exhibit severe limitations regarding lateral stability under critical dynamic regimes and induce roll-induced torsional loading in flexible chain drives. The proposed solution utilizes a twin-tire rear assembly integrated into an articulated suspension mechanism with two degrees of freedom (2 DoF), which reconfigures the geometric stability polygon from a triangle into an isosceles trapezoid. A mathematical model based on tire dynamics and tire slip phenomena demonstrates that introducing a controlled roll stiffness on the rear axle stabilizes the slip angles, ensuring a neutral and predictable steering behavior. Structural validation via finite element analysis (FEA) performed in SOLIDWORKS Simulation on the entire assembly under a conservative combined load scenario (2400 N vertical force shared by the two wheel bearings, 2400 N lateral force, and 1200 N tractive force) indicated a minimum factor of safety of 1.26 on S275N structural steel, confirmed by an eleven-run mesh independence study. Finally, the system’s functionality was experimentally confirmed through the manufacturing and road testing of a full-scale (1:1) demonstrator vehicle powered by an 1129 cc Boxer engine, highlighting a measurable increase in rollover resistance and trouble-free operation of the two-stage chain drive throughout the test program. A numerical evaluation shows that for rear-biased vehicles of the category the proposed axle raises the rollover-related lateral acceleration threshold by up to 54% and replaces the strongly oversteering balance of the single-wheel layout with a near-neutral, tunable one.
Predicting the remaining useful life (RUL) of lithium-ion batteries is essential as it relates to the safety, reliability, and maintenance of both electric vehicles and energy storage systems. While machine learning algorithms have greatly enhanced the accuracy of prediction, a lot of the models available today still have some major flaws including redundant features, poor performance due to non-optimized model settings, and the necessity of the complete degradation process of the system. In this study, we develop a machine learning framework using particle swarm optimization (PSO) to perform hyperparameter tuning and feature optimization to improve lithium-ion battery RUL prediction. The framework allows for the automatic determination of the key degradation indicators, resulting in the optimization of the learning model and enhancement of both the accuracy and interpretability of the predictions. Furthermore, the framework is designed to be employed in an early-cycle learning environment wherein only the first 30% of battery discharge cycles are utilized in the model training. This method is intended to mimic the constraints of real-world applications in which complete lifecycle data is unavailable. Under a properly nested evaluation protocol, the results indicate that the model achieves a mean R2 of 0.354 ± 0.157 and RMSE of 19.13 ± 1.39 using only 30% of degradation cycles, compared to a mean R2 of 0.922 ± 0.014 and RMSE of 18.10 ± 1.11 under full-cycle training across five independent evaluations. While early-cycle prediction shows reduced and more variable explanatory power due to the smaller sample size, the model maintains reasonable error magnitude, supporting its potential utility for early diagnosis of lithium-ion batteries. The effectiveness of the framework is confirmed by feature importance analysis, evaluation of optimization convergence and multi-metric error assessment. In conclusion, the framework offers a highly interpretable and effective solution for predicting lithium-ion battery RUL, in alignment with the emerging data-driven approaches to battery management systems.
As urban rail transit systems enter a stage of high-quality development, accurate short-term passenger flow forecasting has become essential for efficient operation and management. To improve the accuracy and robustness of multi-step short-term passenger flow forecasting under fluctuating demand conditions, this study develops a feature-enhanced Informer framework that embeds Complex Network Representation (CNR) into the Informer architecture to jointly capture both the topological characteristics of the subway network and spatiotemporal passenger flow dynamics. External factors, including subway schedules and land use around stations, are further integrated to enrich the input features. In addition, the ProbSparse self-attention mechanism is adopted to improve long-sequence dependency modeling, thereby enabling efficient multi-step passenger flow forecasting. Experiments were conducted on the Beijing metro passenger flow dataset from January to October 2024 to evaluate the proposed model. The dataset covered 264 stations and was aggregated at 15 min intervals. Based on historical passenger flow and multi-source features, the model predicts passenger flow over multiple future time steps. The overall evaluation metrics were calculated on the test set and averaged over all test samples and observed stations. The experimental results show that, compared with the standard Transformer model, the proposed model reduces the average prediction error by 16.59% on weekdays and 20.48% on weekends while maintaining stable predictive performance during peak hours. Sensitivity analysis and ablation studies are further conducted to evaluate the model performance across different station types and forecasting horizons. The results demonstrate that the proposed model can provide reliable decision support for intelligent urban rail transit operations, including transport capacity scheduling, passenger service improvement, and operating cost reduction.
Emission inventories for high-altitude Andean cities rely on sea-level certification cycles that misrepresent real-world combustion conditions. This study derives altitude-resolved emission factors (EFs) for light-duty gasoline vehicles across 0–4000 m a.s.l. in Ecuador using the purpose-built Andean Ecuador Driving Cycle (aedc), validated against naturalistic data via the Percentile Speed Vector metric (PSVm10; IGS =1.89 vs. IGS =2.30 for the WLTC). Ten vehicles (Euro III–V) were instrumented with OBD-II and portable analysers recording CO, NO, HC, and CO2 at 1 Hz over a four-year campaign (2021–2025; ≈2000 h). K-Means clustering on PSVm10 identified five operating regimes (silhouette ≈0.384). Under dynamically equivalent aedc conditions, NO, CO, and HC all peaked in the 1000–2000 m band (NO: 0.188gkm−1, 6.7× the sea-level value; CO: 4.47gkm−1, +50%; HC: 0.047gkm−1, +292%), fell in the 2000–3000 m band, and partially rebounded above 3000 m (NO: 0.186gkm−1); CO2 instead declined monotonically with altitude (182 to 119gkm−1, −35%), tracking a near-stable-to-slightly-declining fuel consumption (8.56 to 8.11L/100km) consistent with reduced aerodynamic drag at altitude partially offsetting the density penalty. These results show that altitude affects pollutants through distinct, non-monotonic mechanisms rather than a uniform trend, so that single-coefficient altitude corrections introduce systematic bias in Andean emission inventories.
This study investigates the influence of road-induced vertical excitation on air-gap eccentricity in in-wheel permanent magnet synchronous motors. A coupled electromechanical simulation framework is developed by integrating a field-oriented controlled PMSM model, a quarter-vehicle vertical dynamics model, and stochastic road roughness generated according to the ISO 8608. The proposed motor model is validated against experimental data obtained from a dynamometer test bench. Mixed eccentricity conditions are introduced to investigate how road excitation affects air-gap variation and motor current characteristics. A normalized torque-ripple current index is then extracted from the time-domain features of the q-axis current iq, which can be readily calculated from the phase currents and rotor position available in conventional inverter drives without requiring additional sensors. The simulation results reveal that road excitation significantly increases the fluctuation of air-gap eccentricity and amplifies torque-ripple-related current variations compared with no-road conditions. Furthermore, the proposed index increases consistently with eccentricity severity, while rougher road profiles shift the current response toward higher abnormality levels. These findings demonstrate that road–motor coupling has a significant impact on electrical fault signatures and should be considered when developing current-based condition monitoring methods for in-wheel PMSMs. The proposed framework provides a validated basis for evaluating eccentricity faults and supports the development of robust fault diagnosis techniques under realistic vehicle operating conditions.
To address the issues of slow response to preceding vehicles and poor string stability in distributed platoon control of connected and autonomous vehicles (CAVs) under mixed traffic flow, this paper proposes a sliding mode control method based on LSTM trajectory prediction, denoted as LSTM-SMC, within a multi-agent framework. The LSTM model is trained using the HighD naturalistic driving dataset to achieve high-precision prediction of the leader vehicle’s trajectory over a horizon of 3 s, with root mean square errors (RMSE) of 8.52 m in the X-direction and 0.896 m in the Y-direction. The predicted trajectory information is converted into a preview error and embedded directly into the design of the sliding surface, enabling each following vehicle to anticipate disturbances before they propagate. A diminishing preview gain strategy (γ1=0.4, γ2=0.2, γ3=0.1) is employed to suppress error propagation along the platoon, while a saturation function is introduced to eliminate chattering and ensure smooth control inputs. Three simulation scenarios—prescribed leading, HDV (human-driven vehicle) leading, and curved road scenario—are constructed to validate the proposed method against traditional constant time headway (CTH) control, pure sliding mode control (SMC), and LSTM-MPC. Results demonstrate that under extreme conditions, the proposed method reduces the speed RMSE of the 3rd following vehicle by 18.3% compared to CTH and by 39.7% compared to SMC. Under HDV leading conditions, all string stability amplification factors are less than 1, and the position RMSE of the 3rd vehicle is only 5.03 m in the curved road scenario. Compared with LSTM-MPC, the proposed LSTM-SMC achieves comparable tracking accuracy while reducing computational cost by 1.43–3.51×. The proposed method achieves a native integration of prediction and robust control, significantly improving tracking accuracy, string stability, and computational efficiency across diverse operating conditions in mixed traffic flow.
An AI-based methodology was developed for estimating the state-of-health (SOH) of lithium-ion batteries based on secondary operational data and benchmarked with ANN, SVM, RF, and BiLSTM models. The proposed framework was evaluated by using tolerance-based accuracy, Bland–Altman agreement analysis, residual autocorrelation diagnostics, and Cartesian Taylor diagram comparison. The BiLSTM model was the best among the tested models for SOH prediction, with the least prediction error, best agreement with the reference SOH values, and near-white-noise residual behavior. The framework was further extended to Remaining Useful Life (RUL) prediction, where the BiLSTM model showed the most consistent overall performance. We also propose a residual-based anomaly detection as a potential extension of the battery monitoring framework. However, a quantitative evaluation of anomaly detection is out of scope in this study due to the lack of labeled anomaly data in the CALCE dataset. The proposed framework is validated by complementary statistical diagnostics, providing a robust and practical framework for non-intrusive battery health monitoring.
The transport sector plays a significant role in air pollution, and real-world emissions measurements are becoming increasingly important. In this study, emissions from a turbocharged, direct-injection gasoline internal combustion engine (ICE) vehicle and a port fuel injection (PFI) hybrid electric vehicle (HEV) were compared using a portable emissions measurement system (PEMS) under real-world driving conditions. The CO2, CO, NOx, and PN emissions of the two vehicles were measured in urban, rural, and motorway sections. HEV CO2 emissions were ~20% lower than ICE emissions in the entire Real Driving Emissions (RDE) cycle, while in urban operation, they were almost 50% lower. PN emissions were lower for HEV in rural and motorway sections than for ICE, but significant PN peaks occurred during the early urban phase, attributable to the slower engine warm-up of the HEV. Machine learning analysis (Random Forest and Extra Trees Regressor) indicated that coolant temperature was the dominant driver of HEV PN emissions. The results indicate that powertrain characteristics and thermal management strongly influence real-world driving emissions, highlighting their importance for the further development of hybrid vehicles.
Foggy traffic scenes pose significant challenges for object detection because reduced contrast, blurred object boundaries, and the loss of local details weaken discriminative feature representations. These degradations are particularly detrimental to lightweight detectors used in intelligent transportation and vehicle perception systems, where both accuracy and real-time efficiency are required. To address this problem, this paper proposes YOLO-GCM, a lightweight detector-side feature enhancement framework built upon YOLO11n. Instead of relying on an external image dehazing stage, YOLO-GCM improves the internal feature representation of the detector through three complementary modules: a gated additive feature block (GAFB) for adaptive channel-wise feature selection and noise suppression, a context-aware feature enhancement module (CAFEM) for strengthening high-level semantic context, and a multi-scale adaptive fusion (MSAF) module for enhancing cross-scale feature interaction. By integrating these modules into a unified one-stage detector, the proposed method improves detection robustness under low-visibility traffic conditions while maintaining a compact architecture. Experiments on the FoggyCar dataset show that YOLO-GCM achieved 89.81% mAP@0.5 and 67.99% mAP@0.5:0.95, outperforming standard YOLO baselines and dehazing-assisted detection pipelines under a consistent evaluation protocol. Additional evaluation on Foggy Cityscapes further verified the generalization capability of the proposed method under domain shift. The results demonstrate that detector-side feature enhancement provides an effective and efficient alternative to multi-stage dehazing-plus-detection pipelines for foggy traffic object detection. These findings can provide useful guidance for the development of robust and efficient perception modules in roadside monitoring, intelligent transportation systems, and vehicle-assisted driving applications under adverse weather conditions.
Four-wheel-independent-drive electric vehicles are gaining increasing research attention due to their comprehensive dynamic performance. Real-time tire–road friction coefficient information contributes to the development of adaptive control algorithms and active safety control systems for such vehicles. However, traditional tire models widely adopted in existing estimation methods may fail to match practical tire characteristics accurately. Furthermore, lateral velocity serves as a critical state variable for tire–road friction coefficient estimation, whereas existing lateral velocity observers using low-cost inertial measurement unit sensors suffer from degraded estimation performance under complex driving maneuvers. To address the above challenges, this paper proposes a three-stage friction coefficient estimation framework. Firstly, vehicle lateral velocities are estimated via a piecewise gain-scheduled observer using inertial measurement unit measurements. Secondly, tire slip ratios are calculated based on the observed lateral velocities; meanwhile, the longitudinal, lateral and vertical forces of each tire are reconstructed. Lastly, tire force and slip information under combined slip conditions are acquired, and a multilayer perceptron neural network is established to achieve individual tire–road friction coefficient estimation. The simulation results verify the numerical feasibility and preliminary effectiveness of the proposed estimation method under ideal simulation conditions.
