<p>This paper focuses on developing a novel observer, employing a nonlinear decoupling approach to estimate the unknown road input and state variables of a quarter-car suspension system in a laboratory setting. In the proposed method, the effects of the unknown road input, along with the nonlinear characteristics of the elasto-damping elements of the system, are decoupled from the observer equations by selecting appropriate outputs for the suspension system. The superiority of the proposed nonlinear model-based observer is demonstrated through comparison with a recently developed linear observer under the wide range of conditions encountered in real driving. Its efficacy is further validated through experimental implementation on a McPherson suspension test rig. The proposed observer outperforms established nonlinear designs, including the unknown-input extended Kalman filter, extended state observer, and super-twisting observer, by delivering superior accuracy in estimating both unknown road profiles and critical suspension states. Furthermore, its design is simpler to implement, easily extendable to full-car models, and requires no prior assumptions. Leveraging the output of this observer, a new controller for the active suspension system is designed using a continuous predictive control framework. This framework employs fuzzy logic to handle input constraints within the soft computing paradigm. Validation via a co-simulation environment combining Adams and Matlab shows a notable advantage in computational speed for this fuzzy-predictive controller integrated by the proposed decoupling observer over a traditional nonlinear model predictive control, leading to better performance from the overall suspension control system.</p>