Clonal Selection Algorithm Applied to Object Recognition in Mobile Robots

EasyChair Preprint 1279

Abstract

This paper proposes an algorithm for outline recognition in mobile robots, based on Clonal Selection algorithm, a machine learning technique of Artificial Immune Systems. The model is defined from the Industrial process chain point of view, where a robot should recognize object outlines and transport them based on their geometric forms. This detection process should be also rotation and translation invariant. The robot is equipped with color sensors and laser sensors for classi-fying contours. The robot should have the capacity to recognize new objects, classifying those different from the existing ones. The results showed that the Clonal Selection Algorithm in a mobile robot generated antibodies for the correct classification of objects, regardless of their geometrical shape, either defined or undefined geometrical shapes, or even irregular shapes, and including objects with modified contours due to wear and tear, and white noise in the sensors. Therefore, whenever new objects were introduced to the chain process, the robot was successful trained to correctly classify them.

Keyphrases: Artificial Immune Systems, Outline Recognition, clonal selection algorithm, mobile robots

@booklet{EasyChair:1279,
  author    = {Jose Guillermo Guarnizo and Luis Fernando Niño},
  title     = {Clonal Selection Algorithm Applied to Object Recognition in Mobile Robots},
  howpublished = {EasyChair Preprint 1279},
  year      = {EasyChair, 2019}}