Position error prediction using hybrid recurrent neural network algorithm for improvement of pose accuracy of cable driven parallel robots

Abstract

Because cable-driven parallel robots (CDPRs) have lightweight moving parts, CDPRs have been used in various industrial applications requiring high speeds and accelerations. Especially, CDPRs with polymer cables can achieve higher accelerations and speeds compared to those with steel cables. However, they also have some disadvantages, such as a nonlinear creep, a hysteresis, and a short- and long-term recovery. Because these nonlinear characteristics, the accuracy of CDPRs gets worse and worse. In this study, we proposed a hybrid recurrent neural network (H-RNN) to predict nonlinear characteristics of the cable elongation and to solve the problems associated with CDPRs and apply the real-time control. In the algorithm, the long short-term memory (LSTM) algorithm was used to learn the characteristics of the low-frequency data, and the basic RNN learned the features of the high-frequency data. We also confirmed that the cut-off frequency was determined based on the non-operating frequency related to rest time. Also, it yielded more accurate results because the LSTM has a wider effective frequency range. Finally, the learning process was completed by combining these two algorithms. These results made it possible to predict position errors of CDPRs with high accuracy, in which error varies under both while operating and no operation conditions. The H-RNN had a lower root mean square error than both the optimal RNN and the optimal LSTM, so it was effective for controlling systems that have errors across a range of frequencies.