This study is aimed to address the centralized fusion (CF) robust estimation problems for a class of multisensor networked systems with mixed uncertainties, which include coloured noises, same multiplicative noises in system parameter matrices, additive white noise, uncertain noise variances, as well as the one-step random delay and packet dropouts (PD). To deal with this kind of systems, one-step random delay and inconsecutive packet dropouts are described by two Bernoulli distributed random variables. By using the augmentation, de-randomization and fictitious noise techniques, the original system has transformed into an augment model with only uncertain noise variances. In the light of the minimax robust estimation principle, based on the worst-case CF system with conservative upper bounds of uncertain noise variances, the robust CF steady-state Kalman estimators (predictor, filter, and smoother) are presented. By providing a novel approach, consist of non-negative definite matrix method and Lyapunov equation approach, prove the robustness of the CF estimator. Finally, a UPS simulation example illustrates the effectiveness of the proposed CF estimation algorithms.