New adaptive control cuts cable-driven robot error by 34%

Researchers at a Chinese university (Wenbo Gao, Yaoyao Wang, Jiawang Chen, Wenliang Zhang, Hanzhuo Wang) have published a paper on arXiv (2607.26383) presenting a novel control strategy for cable-driven manipulators. These robots, with their inherent nonlinearities and low stiffness, struggle with precise control under time-varying uncertainties and disturbances. The team's approach combines time-delay estimation (TDE) with an adaptive fractional-order nonsingular terminal sliding mode (AFONTSM) controller.

The main contribution is a new nonlinear adaptive law that introduces an adaptive exponential term into the update gain. This design suppresses noise-induced chattering during smooth tracking while preserving or even increasing adaptive gain during trajectory reversal. Experimental results demonstrate significant improvements over baseline methods: RMSE reduced by 34.52% and 31.11% for two joints, ITAE by 33.79% and 32.97%, and integral of squared control torque (ISCT) by 6.69% and 17.77%. Additional payload tests confirm robustness and repeatability, with Lyapunov analysis proving ultimate uniform boundedness of tracking error.