To improve the reliability of the dynamic system including physical and control design, the reliabilityâbased control coâdesign (RBâCCD) problem has been studied to account for the uncertainty stemming from the random physical design. However, when encountering RBâCCD in the sophisticated system in which the dynamic model simulation is timeâconsuming or the state equation is expressed implicitly, the available RBâCCD methods will consume significant computational effort to perform numerous system simulations for the reliability analysis and deterministic optimization. Therefore, this work proposes a Dendrite Netâbased decoupled framework for RBâCCD to alleviate the computational burden. Specifically, the Dendrite (DD) model constructed by the suggested training scheme integrated with an adaptive sampling strategy is used to approximate the state equation in the dynamic system. After that, the sequential optimization and reliability assessment method decouples RBâCCD into the control coâdesign (CCD) problem and timeâdependent reliability assessment problem, which are solved sequentially based on the cheap estimations of DD model, rather than the expensive simulations of the original system. Furthermore, two numerical examples and an engineering example of 3âDOF robot system are applied to demonstrate the feasibility and efficiency of the proposed framework.