TL;DR
This paper introduces a novel coordinated control strategy for autonomous vehicles, specifically in-wheel motor-driven ones, to simultaneously achieve accurate trajectory tracking and maintain vehicle stability, even under challenging conditions like low-adhesion roads. The approach combines adaptive trajectory tracking with a fuzzy-enhanced sliding-mode controller for stability, dynamically adjusting their interplay using a normalized stability index.
What Problem Does This Paper Address?
Intelligent driving systems aim to enhance safety and efficiency, but a critical challenge arises when autonomous vehicles operate under extreme conditions. These include sudden braking, low-friction surfaces (like ice or snow), or high-speed maneuvers. In such scenarios, the vehicle's tendency to become unstable can severely compromise its ability to accurately follow a planned path. Traditional control systems often struggle with the inherent conflict between trajectory tracking (keeping the car on the desired path) and stability control (preventing spins or loss of control). This paper tackles this by developing a unified strategy that intelligently balances these two crucial aspects for in-wheel motor-driven vehicles, which offer greater control flexibility.
Methods
The proposed solution is a hierarchical, yet integrated, control architecture with several key innovations:
- Adaptive Trajectory Tracking: Instead of relying on fixed vehicle parameters, this controller uses an adaptive Model Predictive Control (MPC) approach. A crucial element is the real-time estimation of tire cornering stiffness using a Forgetting Factor Recursive Least Squares (FFRLS) algorithm. This allows the controller to adapt to changing road conditions and tire behavior, improving tracking accuracy.
- Fuzzy-Enhanced Stability Control: For stability, an integral sliding-mode control (SMC) is employed to generate the necessary direct yaw moment. A significant challenge with SMC is "chattering" (high-frequency oscillations). This is mitigated by a fuzzy controller that adaptively adjusts the SMC's reaching law coefficient, smoothing out the control output.
- Coordinated Control Mechanism: The core of the coordination lies in a normalized stability index. This index quantifies the vehicle's instability level. Based on this index, adaptive weight coefficients dynamically adjust the priority between trajectory tracking and stability control. When instability is high, stability control takes precedence; when stable, trajectory tracking is emphasized.
Datasets
Not discussed in the paper. The evaluation relies on hardware-in-the-loop (HIL) simulations using CarSim and MATLAB/Simulink.
Key Insights and Ideas
- Tire Cornering Stiffness is Key: The paper highlights that accurately estimating tire cornering stiffness in real-time is fundamental for robust trajectory tracking, especially when tire forces approach saturation.
- Fuzzy Logic for SMC Smoothing: Using fuzzy logic to tune SMC parameters is an effective way to combat chattering, a common issue that can degrade control performance and comfort.
- Dynamic Prioritization is Crucial: The normalized stability index and adaptive weighting provide a sophisticated way to manage the trade-off between tracking and stability, ensuring the vehicle prioritizes safety when needed without overly compromising its ability to follow the path.
- In-Wheel Motors Offer an Advantage: The inherent flexibility of in-wheel motors in independently controlling each wheel's torque is leveraged to enhance both stability and tracking.
Key Results
The proposed coordinated control strategy was validated through hardware-in-the-loop simulations under various scenarios:
- Low-Adhesion Road (Case 1): The adaptive MPC controller showed improved trajectory tracking compared to a standard MPC, with better handling of lateral forces and cornering stiffness variations.
- Emergency Obstacle Avoidance (Case 2): The proposed controller (Controller 3) significantly outperformed a trajectory-tracking-only controller (Controller 1) and a less integrated coordinated controller (Controller 2). It demonstrated smaller lateral and heading errors, reduced yaw rate and sideslip angle peaks, and better overall performance in lane-changing maneuvers, especially under high-adhesion conditions.
- Transitioning Road Surfaces (Case 3): The strategy showed robustness to sudden changes in road adhesion, maintaining stability and accurate trajectory tracking where other controllers failed. The adaptive nature of the controller, particularly the real-time stiffness estimation and adaptive weighting, was crucial here.
Limitations and Future Work
- Computational Intensity: The paper notes that integrated control architectures can be computationally intensive, potentially impacting real-time performance. While their hierarchical approach aims to mitigate this, further optimization might be needed for extremely complex scenarios.
- Parameter Tuning: The effectiveness of the fuzzy controller and the stability index relies on careful tuning of parameters (e.g., forgetting factor, fuzzy membership functions, stability index thresholds).
- Future Work: The paper implicitly suggests that further exploration of adaptive control techniques, more sophisticated vehicle models, and real-world testing would be valuable next steps. The adaptive adjustment of weight coefficients and the fuzzy logic for chattering reduction are areas that could be further refined.
Summary updated