Disturbance-resilient model predictive control for active suspension systems with perception errors in road preview information

The vehicle suspension system is a critical component of automobiles, primarily designed to mitigate the impact of uneven road surfaces, thereby enhancing ride comfort, while also influencing handling by connecting the chassis to the wheels. With advancements in sensor and automation technologies, the concept of preview-active suspension has emerged. This technology utilizes forward-facing sensors to gather road information, allowing for proactive adjustments to the suspension system to optimize both comfort and handling. Although forward-facing sensors help reduce uncertainties in road profiles, sensor noise and environmental disturbances can introduce perceptual errors, which may lead to instability and the loss of recursive feasibility in the system, resulting in pathological behavior. To address these challenges, this paper proposes a Disturbance-Resilient Model Predictive Control (DR-MPC) based active suspension controller to mitigate the pathological issues caused by preview perception errors. By modifying the standard terminal cost function of MPC and introducing constraints on perceptual errors, the proposed controller effectively suppresses disturbances caused by these errors. Stability and recursive feasibility of the controller are rigorously proven. Finally, the effectiveness of the proposed algorithm is validated through real-vehicle road data collection and hardware-in-the-loop (HIL) testing.