Inverse kinematic problem of a cable-driven parallel robot (CDPR) is a straightforward problem for massless inextensible cables. However, the problem becomes a non-linear kineto-static problem with the consideration of cable mass and elasticity. This problem is conventionally solved using numerical methods. These methods are iterative in nature, have slow convergence speed, and are highly dependent on initial values, which makes them difficult to be used in real-time applications. Therefore, this paper proposes a deep neural network-based approach to obtain the inverse kineto-static (IKS) solution of CDPR considering cable mass and elasticity. The training dataset has been obtained from IKS solutions attained by using numerical methods. Then, appropriate parameters as well as training algorithm is utilized for off-line training of the network. The performance of the proposed approach is validated by the simulation results for a redundantly constrained planar CDPR. From the simulation results, it has been observed that the proposed approach shows an 83% reduction in computational time in comparison to conventional numerical methods. In addition, the proposed neural network-based approach has performed satisfactorily with a minor error in the IKS solution. Therefore, the proposed approach can be employed for real-time applications due to its low and bounded computational time.
Neural Network-Based Inverse Kineto-Static Analysis of Cable-Driven Parallel Robot Considering Cable Mass and Elasticity
Mechan. Machine Science
International Conference on Cable-Driven Parallel Robots ; 2021 July 07, 2021 - July 09, 2021
01.06.2021
13 pages
Aufsatz/Kapitel (Buch)
Elektronische Ressource
Englisch
An Inverse Kinematic Code for Cable-Driven Parallel Robots Considering Cable Sagging and Pulleys
Fraunhofer Publica | 2020
|The Static Characteristic Analysis of the Long-Span Cable-Driven Parallel Robot
Springer Verlag | 2024
|Elasto-Static Model and Accuracy Analysis of a Large Deployable Cable-Driven Parallel Robot
Springer Verlag | 2023
|