| This thesis focuses on the application and theoretical developments of model predictive control techniques for the distributed control and safe coordination of autonomous vehicles, specially those with nonholonomic dynamic models. We consider in particular the problem of safe trajectory tracking for a group of interacting autonomous nonholonomic vehicles. In addressing this problem, new model predictive control formulations, new model predictive control stability results, and novel distributed coordination schemes are presented. Proposed model predictive control stability results are associated with two new formulations: constructive model predictive control and modified contractive model predicted control. In these schemes, we provide different types of stability guarantees by incorporating extra stability constraints in the proposed optimization formulations. These results are then used to develop a decentralized coordination control policy that addresses a modified version of the multi-vehicle safe trajectory tracking problem and considers overall system stability and safety issues by incorporating elements from hybrid system automata, multiple Lyapunov functions, and modified avoidance functions related to avoidance control. Finally, a novel model predictive control coordination scheme for spatially distributed systems that are fully cooperative (based on a 2-layer formulation) and incorporate convergent over-approximations of the maximum function in the objective functions is presented. The novel approach is illustrated by formulating the multi-vehicle safe trajectory tracking problem in this framework and providing a comparative study for two different scenarios. |
