Control y planificación óptima de sistemas extendido a plataformas no lineales e inestables

Abstract

The motivation of this Thesis has been to take a step further in the research of Optimal Control techniques. I was not the first researcher to work on this technique and once, we have seen the achieved results, this development will be continued. The main ability of this algorithm is not the implementation of an efficient controller, but letting us make an adaption to possible changes in the plant and obtain some non-invasive metrics to evaluate its evolution. Furthermore, this controller can be also applied to path planning of mobile systems in which an optimal trajectory is traced to a goal considering all elements around. This capability is especially useful to optimize some specific parameters such as its autonomy, travel time, energy consumption or comfort during the travel. Finally, one of the most important issues of the controller that has been taken into account in this Thesis is its capability to be implemented in real world considering computational constraints. In order to reach this, MultiScale technique has been developed together with CACM-RL* algorithm