Comparison of neural network architectures for the modeling of robot inverse kinematics

摘要

Describes the use of neural networks to model the inverse kinematics of robot manipulators, including a redundant manipulator The use of multiple cooperating networks for the overall modeling of inverse kinematics was explored. A variety of network architectures was used, and their performance was compared. Neural networks were also used to train robots in specified obstacle-avoidance trajectories.