Applied Artificial Neural Networks and Genetic Algorithms in Simulation Strategy for Trajectory in Collaborative Robotic

Abstract

This work proposes a solution for collaborative robotics, presenting some precautions with this new perspective of robotics and norm, trajectory planning and observed singularities. Also, it compares techniques for solving three degrees-of-freedom (DOF) robotic manipulator inverse kinematics based on genetic algorithms (GAs) and artificial neural networks (ANNs). In addition, a decision tree was included to increase the arm motion's safety when an object appears in its trajectory, simulating an environment in which collaborative robots work side by side with human beings. Fifth-order polynomials were also compared in trajectory planning, and the analysis showed that the fifth-order polynomial presented a trajectory solution. According to the results obtained from ANN and GA performance, the efficacy of the proposed methodology was demonstrated. The proposed research study demonstrates unique potential, especially in initial 3D tests, providing robust results with a time-sufficient solution. To conclude this scientific investigation, conclusions and possible future developments of this research are summarized.