I. Introduction
Motion planning and autonomous vehicle control problems in dynamic and uncertain environments are of paramount importance in several human applications where guaranteed real-time feasibility, safety against uncertainties and disturbances are required. Despite extensive research, this problem still represents a real challenge that becomes more complex by taking into account the limited perception abilities and computational power resources of the robot. Though the obstacle avoidance issue is a key component of the safe navigation, most of the algorithms proposed in the literature only addresses the path planning aspect by leaving out the control phase (see [15], [21], and the references therein) or partially tackled by considering specific kinematic vehicle descriptions. In particular, probabilistic roadmap methods are considered in [16] and the proposed algorithm creates a robust roadmap in the preprocessing phase using the observation that the behavior of the moving obstacles is often not unconstrained but restricted to prespecified areas. In [18], the so-called partial motion planner mechanism is capable of generating safe paths although the high computational burdens could lead to violate the real-time constraints under which the robot must take a decision. In [21], the path planning problem under nonholonomic constraints is addressed using the so-called Follow the Gap Method. Along these lines is also the recent contribution [20], where the attention has been focused on the formal definition of a navigation algorithm within dynamic cluttered environments.