This paper addresses the trajectory tracking control problem of robotic wheelchairs in the presence of modeling uncertainties. The controller has been designed using position and angular measurements. A global ultra-model, or simplified model achieved from flatness considerations is proposed first. This model highly reduces the design complexity of the state estimation and the output feedback control tasks since it groups, as an unknown time-varying disturbance, both the combined effects of all uncertain state-dependent (i.e., endogenous) nonlinearities and those of external (i.e., exogenous) perturbation inputs which are present in the input-to-flat output model of the system. An extended linear high-gain observer, or Generalized Proportional Integral (GPI) observer, is then developed for the simultaneous, though approximate, state and disturbance estimation. The proposed feedback controller combines the global ultra-model and the GPI observers to conform an active disturbance rejection, or disturbance accommodation, control scheme. The simulation results presented in the paper show that the proposed method has a very good tracking performance and robustness in the presence of system uncertainties, external disturbances and noisy corruptions.