Robust Iterative Learning for Collaborative Road Profile Estimation and Active Suspension Control in Connected Vehicles

TL;DR

This work tackles robust road-profile estimation for active suspension in connected vehicles by introducing a collaborative, multi-vehicle framework that crowdsources data to mitigate vehicle heterogeneity and modeling uncertainties. It combines disturbance observer-based estimation with cascaded iterative learning control, establishing a recursive relation $e_{d,j}=T_{e_{1,j}}\{e_{d,j-1}\}+T_{e_{2,j}}\{d_{f,j-1}\}$ and a theorem that prescribes learning filters to achieve error reduction by a factor $\alpha$ per iteration. The proposed filters are $L_{1,j}=\hat{G}_{d,j-1}^{-1}[\alpha(1-\hat{\Omega}_{j-1})-\eta_j(1-\hat{\Omega}_j)]$ and $L_{2,j}=\alpha+L_{1,j}\hat{G}_{f,j-1}$, ensuring $||e_{d,j}||\approx \alpha||e_{d,j-1}||$ under reasonable small-uncertainty assumptions. Numerical results on 90 dynamically varying agents across sinusoidal and Type-C road profiles demonstrate substantial RMSE reductions (e.g., from ~10.25 mm to ~1.48 mm and ~0.98 mm), validating the approach as a practical add-on for active suspension in connected-vehicle systems.

Abstract

This paper presents the development of a new collaborative road profile estimation and active suspension control framework in connected vehicles, where participating vehicles iteratively refine the road profile estimation and enhance suspension control performance through an iterative learning scheme. Specifically, we develop a robust iterative learning approach to tackle the heterogeneity and model uncertainties in participating vehicles, which are important for practical implementations. In addition, the framework can be adopted as an add-on system to augment existing suspension control schemes. Comprehensive numerical studies are performed to evaluate and validate the proposed framework.

Figures (7)

• Figure 1: Overall learning schematic along with a quarter car model used for one of the agents
• Figure 2: (a) Original block diagram of the system (b) Equivalent block diagram after block diagram manipulation
• Figure 3: Block diagram of the system along with learning framework for Agent$\#(j)$
• Figure 4: Bode plots of all the agents for transfer functions from (a) active suspension force to sprung mass displacement (i.e. $P_j(1)$) and (b) from road profile to sprung mass displacement (i.e. $P_j(2)$)
• Figure 5: Road profiles introduced