Comparação de métodos de otimização Local e Meta-Heurístico para seguimento de caminho do manipulador robótico KUKA KR 6 R700 sixx

Abstract

Performing the inverse kinematics of a robotic manipulator is always a challenge, especially when it comes to several degrees of freedom. Nevertheless, there are already many methods proposed to solve this issue. In this work, the comparison of three methods for solving inverse kinematics is performed, these being local optimization methods, such as the Inverse Jacobian, Levenberg-Marquardt and a global optimization metaheuristic method, the Particle Swarm Optimization (PSO). The algorithms were programmed in C++, which were applied into Gazebo software to performe a 3D simulation of the robot motions. To create an interface between the real system and the simulation environment, the ROS framework was used. In the tests performed, all results demonstrated high accuracy, with the Inverse Jacobian and the Levenberg-Marquardt standing out.