Forward Kinematics Solution of Cable Robot based on Neural Network and L-M Algorithm
- 12 March 2021
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. 5, 2519-2524
Abstract
Cable-driven parallel robots (CDPRs) have been applied very well because of its advantages. Like other parallel robots, the forward kinematic solution of the parallel robot driven by cable is usually faced with complex nonlinear equations, and its solution is always a difficult problem. The traditional numerical iteration method is heavily dependent on the selection of initial values, which is easy to fall into local convergence, leading to the result of calculation is not necessarily the ideal spatial pose. The neural network has good nonlinear fitting ability and can give correct spatial pose, but the solution accuracy needs to be improved. The convergence and precision of the forward kinematic solution of the CDPRs is solved by the improved algorithm combining neural network and Levenberg- Marquardt algorithm.Keywords
Funding Information
- National Natural Science Foundation of China
This publication has 7 references indexed in Scilit:
- Design and Kinematic Analysis of the Novel Almost Spherical Parallel Mechanism Active AnkleJournal of Intelligent & Robotic Systems, 2018
- Nonlinear Electrical Impedance Tomography Reconstruction Using Artificial Neural Networks and Particle Swarm OptimizationIEEE Transactions on Magnetics, 2015
- Large-scale 3D printing with a cable-suspended robotAdditive Manufacturing, 2015
- Neural Network Solution for Forward Kinematics Problem of Cable RobotsJournal of Intelligent & Robotic Systems, 2010
- Dynamic modeling and active control of a cable-suspended parallel robotMechatronics, 2008
- Artificial-Neural-Network-Based Sensorless Nonlinear Control of Induction MotorsIEEE Transactions on Energy Conversion, 2005
- An algorithm for solving the direct kinematics of general Stewart-Gough platformsMechanism and Machine Theory, 1996