Proceedings of the
9th International Conference of Asian Society for Precision Engineering and Nanotechnology (ASPEN2022)
15 – 18 November 2022, Singapore

Inverse Kinematics Selection Algorithm for 6-DOF Manipulator Based on Neural Network to Pass Through the Singular Point

Daiki Katoa, Naoki Maeda, Ayumu Takeuchi, Toshiki Hirogaki and Eiichi Aoyama

Department of Mechanical Engineering, Doshisha University, 1-3, Tatara Miyakodani, Kyotanabe, Kyoto, Japan


The transformation of a robot's end-effector trajectory into each joint angle command is an inverse kinematics problem, which is the most fundamental problem in robot control systems under an offline teaching. This study developed a neural network-based algorithm to avoid the rapid rotation of some joints when a six-degree-of-freedom manipulator passes through the vicinity of a singular point, which is one of the most challenging problems in achieving offline teaching technology in industrial robotics. We defined the vicinity of a singular point as the extent when the speed of the servomotor is more than the maximum rotation speed and constructed the inverse kinematics model using an artificial neural network (ANN). Conventional methods can only solve for continuous functions, and ANN intentionally defines a discontinuous function by limiting the training data to suppress rapid rotations. The results show that the proposed method enables a manipulator to pass through the vicinity of the singular points without decreasing trajectory and velocity accuracy.

Keywords: Robotics, Singularity, Inverse kinematics, Neural network.

PDF Download