Desarrollo de un toolbox para resolver trayectorias de robots manipuladores mediante entrenamiento basado en algoritmos genéticos

Abstract

This project has the objective of analyzing the use of a computational intelligence method such as genetic algorithms to solve the inverse kinematics and plan the trajectory of manipulator robots and based on it to develop a toolbox to visualize and interact with various types and configurations of robots. The toolbox was developed in MATLAB software that has a great capacity to handle matrices and also has tools and functions included for artificial intelligence, the GUIDE space also allows to design the HMI which has been done based on the Gedis guide which gives us parameters to make an interface that is suitable, The toolbox developed allows to solve the inverse kinematics through genetic algorithms and obtain the desired trajectories in a manipulator robot and also to visualize the movement of the links to choose the robot that best suits an application and even design a robot by changing kinematic parameters such as degrees of freedom, The toolbox is an academic tool that allows to observe the behavior of the robot and simulate its movement, it is quite useful for teaching industrial robotics showing in a graphical interface the main parameters that are taken into account in the kinematic aspect also has a database that allows to simulate the main industrial robots in the market and an interface that allows to connect a robot through serial communication. Finally, a series of tests have been carried out, analyzing the accuracy and repeatability of the algorithm in different models of industrial manipulator robots, including tests that validate the usefulness of the genetic algorithms method in redundant robots.