A novel anthropomorphic speed decision and active suspension co-control method considering road roughness for autonomous vehicles

Abstract

To address the conflict between ride comfort and traffic efficiency for autonomous vehicles traversing uneven roads (e.g. speed bumps and potholes), this paper proposes an integrated speed–suspension co-optimization framework that explicitly considers driving styles and perceptual uncertainty. First, addressing the challenge of obtaining precise environmental parameters in real-world scenarios, a Dual-Layer Probabilistic Perception Model and a Chance-Constrained Robust Decision Mechanism are established. By incorporating style-based cognitive bias and stochastic sensory noise, this approach transforms deterministic optimization into a robust planning process resilient to perception errors. Second, by incorporating Kinetic-Energy-Coupled Prospect Theory and a data-driven parameter initialization method (K-means), an anthropomorphic multi-objective speed planning model is developed to accurately reproduce the non-rational risk perception characteristics of conservative, balanced, and aggressive driving styles. Furthermore, a Speed-Aware Preview Model Predictive Control (MPC) strategy is designed for the active suspension system. This mechanism dynamically adjusts the weights of future road disturbances based on real-time vehicle speed, achieving deep coupling between upper-level longitudinal speed decision-making and lower-level vertical suspension control. Co-simulation results demonstrate that the proposed strategy effectively adapts to diverse driving styles and, compared with conventional decoupled methods, significantly improves overall ride comfort and safety under various road conditions.