Desarrollo de algoritmo para detección y comando de robots humanoides en tareas de recolección

Abstract

This article presents an algorithm which commands a group of Bioloid humanoid robots in order to organize them around an object of interest, previously detected by an external vision system. The robots form a Multi-Agent System (MAS) oriented towards cooperative gathering tasks. Development of the MAS, as well as each of the organization algorithm’s components and simulation inside a virtual environment are all detailed. The algorithm is subdivided in two dedicated threads: one of which handles machine vision (filtering, contour detection and classification achieved through EmguCV libraries) and operational space calculations, and another which operates ZigBee wireless communication with the robots. Furthermore, the robots possess their own embedded code which enables them to translate a sequence of received instructions into gait patterns which allow them to move towards the object of interest. Total execution time for the gathering task is chosen as the global performance measure to evaluate.

Keywords

• Computer simulation
• Humanoid robots
• Multi-agent systems
• Robot control
• Robot vision systems