Nuevas aportaciones en algoritmos de planificación para la ejecución de maniobras en robots autónomos no holónomos

Abstract

The starting point of this research was merging the advantages of the RRT with the restricted maneuverings. This new planner has become better than the RRT that takes into account the constraints, and faster in one magnitude order. A new method to obtain Voronoi diagrams called Vodec has been developed. It uses a non-euclidean metric to split the problem in separate cells that are obtained from a Vertical decomposition. It is specially useful in dynamic scenarios and planning tasks. The RRT algorithm provides with different solutions over the same problem. This advantage can be used to generate a set of solutions and select the best. The multicriteria decision methods allow an easy way to do it. An application is added here. An adaptation of the new planner to scenarios with multiple stop points and with wireless sensor networks that can warn about new obstacles, called distributed RRT, is shown.