Implementação de redes neurais artificiais utilizando FPGA para aplicações de controle inteligente em robótica

Abstract

In the following work project deals with the elaboration of a Artificial neural network, Multi-layers feedforward, described in hardware language VHDL for the application in mobile robotics, since this elaboration as a block system, until the fine adjustments like the initialization of the ideal weights for the problem. Mathematical auxiliary software was used for the necessary adjustments in the database, initial weights and learning rate. The training algorithm described in VHDL, the user/hardware interface and the overall system design is left for future work.