Um algoritmo bioinspirado para o processo de tomada de decisão e sua aplicação a robôs móveis

Abstract

Bio-inspired algorithms are mathematical tools based on principles of evolutionary nature and animal behavior for the development of new and robust techniques that aid in different areas of technology, such as engineering and robotics. The development of bio-inspired algorithms and implementation to mobile robots has been widely used in science to solve problems both in robotics and in behavioral animal studies. This work aimed to the use of 𝜺 -greedy for the decision-making process in a mobile robot and to compare its behavior with the Betta splendens fish . The experimental design, the components selection, robot assembly, calibration of sensors and the necessary test routines for the validation of the algorithm were performed. The operation of 𝜺 -greedy was validated. The comparison with previous work performed at the Laboratory of Manufacture by the Roboteam in partnership with the Fish Laboratory showed that the algorithm was properly implemented to the mobile robot, making it able to learn from past experiences and make more advantageous decisions. After validation, the parameter epsilon of 0.4103 was determined and most advantageous in the decision-making process of the mobile robot. This work generated the construction of a mobile robot, validated the use of 𝜺 -greedy to mobile robots and simulated efficiently the computational experiments already performed with the algorithm in the decision making of the fish.