Selección óptima de parámetros para algoritmos de detección de obstáculos con visión monocular.

One of the most important task in the field of mobile robotics is obstacle detection. To solve this task, computer vision has been used often, especially monocular vision. This is due to the inherently complexity of stereo vision systems and the increasing development of the research that uses a single camera to detect obstacles. Computer vision and image processing algorithms for obstacles detection require multiple parameters that need to be adjusted to work efficiently according to the characteristics of the robot and the conditions of the environment in which it operates. In the present work, a method for the optimum selection of the parameters of this kind of algorithm for a certain environment is proposed. To achieve that, the obstacle detection problem was modeled as an optimization problem. Besides, an explanation of two of these algorithms based on monocular vision are given. These are used for the validation of the given method. For the solution of the given problem, obtained results with different metaheuristics are included. Finally, the obtained results from using these techniques in different environments are compared. [ABSTRACT FROM AUTHOR]