masterThesis
Um novo método de planejamento de caminho para robôs baseado em espuma probabilística
Fecha
2016-12-16Registro en:
SILVEIRA, Yuri Sarmento. Um novo método de planejamento de caminho para robôs baseado em espuma probabilística. 2016. 62f. Dissertação (Mestrado em Engenharia Mecatrônica) - Centro de Tecnologia, Universidade Federal do Rio Grande do Norte, Natal, 2016.
Autor
Silveira, Yuri Sarmento
Resumen
Path planning is a well studied problem in robotics. The capability of analyzing the
environment and defining a sequence of actions that leads a robot from an initial location
to a final desired location, without colliding with obstacles, is a fundamental ability when
creating autonomous robotic systems that can perform various functions. In order to contextualize the theme addressed in this work, a succinct study of the state-of-the-art on path planning for autonomous robotic systems is presented. Each planning method has its own strategy to explore the ambient and plan the path. In
this dissertation, a new robot path planning method is proposed. In the proposed method,
the ambient free space is partially covered by a set called Random Foam, composed of
the union of overlapping convex subsets called Bubbles. Starting from the initial robot localization, new bubbles are randomly created on the
surface of the foam, that propagates through the free space, with a behavior similar to a
wave front propagation, generating a search tree that grows until reaching the desired final
robot localization. In this way, it is possible to find a sequence of concatenated bubbles,
called Rosary, connecting the desired final localization to the initial localization of the
robot. A valid path contained in the maneuvering space defined by the rosary can be
easily found. In the proposed method, the search process is guided by only two parameters. Tuning
criteria for these parameters are studied and presented in this work. In order to validate the proposed path planning method, its performance was evaluated through computer simulations of different case studies.