Por favor, use este identificador para citar o enlazar este ítem:
http://inaoe.repositorioinstitucional.mx/jspui/handle/1009/2090| Robotic motion path generation method based on homotopy continuation for multidimensional workspaces | |
| Gerardo Díaz-Arango | |
| Luis Hernández Martínez Héctor Vázquez Leal | |
| Acceso Abierto | |
| Atribución-NoComercial-SinDerivadas | |
| Homotopy Robotic motion planning Multidimensional workspaces | |
| Motion planning is the process that every autonomous agent (robot arm, mobile robot, video game avatar, molecular model, among others) must develop, this consists on calculating a sequence of movements that represent the continuous displacement from an initial position to a goal position. In principle, path planning is a purely geometric procedure in which an autonomous robot must find a collision-free path without considering dynamics and other differential constrains. The present work is devoted to formulating a planner algorithm based on homotopy continuation methods which can deal with most important issues present in robotic motion planning. These issues are; a) workspace model type and number of dimension representation, b) large number of obstacles, c) narrow corridors, d) memory and computing time spend and e) completeness. Furthermore, an additional topic is treated; multiple-goal planning for multidimensional configuration spaces with static obstacles. The main contributions of the present thesis are; first, a multidimensional technique to enhance the convergence of spherical tracking and correct the reversal phenomenon, named double spherical racking (DST). Second, an automatic criterion to assign the repulsion parameter for obstacles in the homotopic path planning method (HPPM) for any 2D workspace is proposed. Third, HPPM is reformulated by using a piecewise linear approach in order to generate a multiple-goal planner. This variation also is used to perform problems in multidimensional workspace (from two to five dimensions). Fourth, four workspace representations for HPPM: i) Algebraic models for multidimensional space by using mathematical representations closed curves and surfaces, ii) Polygonal and polyhedral models, iii) Non-convex polygons and iv) Bitmaps by using piecewise linear approach (in this work only apply for 2D problems) are characterized. The effectiveness of our proposed planner to obtain feasible paths for sceneries with issues, restrictions and workspace models mentioned previously are validated by using numerical simulations. The performance of the proposed homotopic planner is compared with the most employed sampling based planner algorithms (SBP) for the metrics: execution time, length of path, percentage of successful runs and memory consumption. The results of the cases of study show that the homotopic planner is the best option to implement on embedded system with limited resources (processor and memory). | |
| Instituto Nacional de Astrofísica, Óptica y Electrónica | |
| 2018 | |
| Tesis de doctorado | |
| Inglés | |
| Estudiantes Investigadores Público en general | |
| Díaz Arango, Gerardo Ulises, (2018). Robotic motion path generation method based on homotopy continuation for multidimensional workspaces, Tesis de Doctorado, Instituto Nacional de Astrofísica, Óptica y Electrónica. | |
| ELECTRÓNICA | |
| Versión aceptada | |
| acceptedVersion - Versión aceptada | |
| Aparece en las colecciones: | Doctorado en Electrónica |
Cargar archivos:
| Fichero | Tamaño | Formato | |
|---|---|---|---|
| DiazArGU.pdf | 15.03 MB | Adobe PDF | Visualizar/Abrir |