This paper presents a new technique for tracking-error model-based Parallel Distributed Compensation (PDC) control for non-holonomic vehicles. We take into account the delay in the outputs (measurements) in the case of a constant delay. Briefly, this technique consists of rewriting the kinematic error model of the mobile robot tracking problem into a TS fuzzy representation and finding a stabilizing controller by solving LMI conditions for the tracking-error model as it is presented in [1]. The state variables are estimated by a TS fuzzy predictor observer in order to compensate the delay produced by the sensor (camera in our case) and the stability of the observer is proved by solving linear matrix inequalities (LMI) conditions. To illustrate the efficiency of the proposed approach a comparison between the TS fuzzy observer and the nonlinear predictor observer is shown. For this study the reference trajectory is a C3 Bezier curve. All experiments are implemented on simulation.