Ponente
Descripción
This project involves the modeling and control of a two-degree-of-freedom Cartesian robot, incorporating the phenomenon of static friction to enable its parametric identification using convolutional neural networks (CNNs).
The objective is to design and train a CNN capable of identifying a total of ten dynamic parameters present in the Cartesian robot. This aims to compensate for the nonlinear disturbances commonly found in this type of system, while also contributing to the development of an experimental platform for the Master's Program in Electronic Sciences.
A novel approach is proposed to generate images using signals from the robot—such as position, velocity, acceleration, and torque—combined with a set of input dynamic parameters, in order to build the image dataset required for training the neural network.
It is important to note that for this proposal to function correctly, a dynamic model of the robot (including the physical phenomenon of friction) must be simulated in state space using MATLAB.
