Learning by Demonstration applied to a Robotic Arm using Paraconsistent Artificial Neural Cell Paulino Machado Gomes, Mauricio Conceição Mario, João Inácio da Silva Filho, Leonardo do Espirito Santo, Rodrigo Silvério da Silveira, Cláudio Luís Magalhães Fernandes

Abstract

The Annotated Paraconsistent Logic - LPA is a non-classical logic that is based on concepts that allow, under certain conditions, to accept the contradiction in its foundations, without invalidating the conclusions. In this work, an algorithm called Paraconsistent Artificial Neural Cell of Learning (PAN Celll), which was created from equations based on LPA Logic, is used. With standardized signals repeatedly applied to the PAN Celll input, it is possible to gradually store this information, increasing or decreasing its response level at the output with asymptotic variation and controlled by a Learning Factor (lF). A set of five PAN Cellls was implemented in an ATmega 328p microcontroller, forming a learning Paraconsistent Artificial Neural Network (PANNet) and several tests were carried out to validate its operation acting in learning by demonstration (LfD) in a Robot Manipulator. The results of comparative studies showed that PAN Celll has dynamic properties capable of acting, both in the learning process by demonstration and in the imitation process.