I. Introduction
In several applications, autonomous mobile robots are required to operate in uncertain environments avoiding obstacles along the path [1]. Despite extensive research [2] this subject still represents a complex challenge, particularly when robot's inaccurate perception capabilities, limited computational resources, actuation constraints and under-actuation phenomena are considered. For example, the control of wheeled differential-drive robots is enlightening because of the presence of non-holonomic constraints [3]. Indeed, it has been proved that a pure-state feedback controller capable of asymptotic stabilizing fixed configurations does not exist [4]. As long as the obstacle avoidance and motion planning issues are concerned, the available solutions range from optimization-based methods, reachability analysis to graph search theory [5]. Besides these approaches, also Model Predictive Control (MPC) strategies have been successfully applied thanks to their capability of jointly addressing performance and collision avoidance specifications. In [6] the authors propose a nonlinear MPC scheme for ensuring tracking capabilities in the presence of slippery roads for safe and pre-assigned reference trajectories, while in [7] cluttered obstacle scenarios are taken into consideration. In [8], operating scenarios with unexpected obstacles occurrences are of interest. There, a two-mode nonlinear MPC is developed whose one of the main feature relies on the capability of confining the regulated vehicle trajectory within an invariant safe region. In [9], set-theoretic ideas and sum of squares decomposition techniques are used to develop a receding horizon control scheme for constrained wheeled vehicles described by polynomial models.