I. Introduction
To obtain a fully automated transportation system, different issues must be addressed [1]. Each vehicle must be able to track a reference signal and reach the desired goal [2], [3], formation control tools are needed to guarantee coordination among vehicles [4], [5], and absence of collisions must be ensured [6]. Of interest for this article are control solutions capable of addressing the collision-free reference tracking control problem for multiple unmanned vehicles (MUVs) moving in a 2-D planar environment. In [5] and [7]–[12], different multiagent strategies have been proposed to address the collision avoidance problem. Such approaches, although appealing, cannot be straightforwardly used to take into account the simultaneous presence of state and input constraints and exogenous disturbances. In [13], collision-free movements and deadlock avoidance for multirobot systems are achieved by defining a proper quadratic programming (QP) problem based on a mixture of relaxed control barrier functions, hybrid braking controllers, and consistent perturbations. In [14], the collision avoidance problem for a group of vehicles is formulated as a mixed-integer linear programming (MILP) problem, where each vehicle has an a priori known fixed number of possible trajectories. Along similar lines is also the work in [15], where the exponential complexity of the obtained Hamilton–Jacobi (HJ) and MILP problems is reduced using a combinational technique based on HJ reachability and MILP programming concepts. In [16], a robust decentralized model-predictive controller (DMPC) for a team of unmanned vehicles (UVs) is proposed. In particular, absence of collisions is modeled as coupling constraints, and the resulting centralized MILP optimization problem is recast into smaller size MILP problems, sequentially solved by the vehicles. Such a solution, at the cost of increasing communication requirements among the vehicles and conservativeness, is proved to better scale with the number of vehicles with respect to the centralized counterpart. In [17], it has been shown that in structured environments, with a finite and a priori known number of trajectories, the intervehicle communication demand of decentralized solutions can be reduced by resorting to a centralized entity in charge of coordinating the vehicles trajectories.