ABSTRACT
Decoupling algorithms can be classified into linear or nonlinear decoupling categories. It is necessary to have an evaluation indicator to judge the different decoupling algorithms of multi-component force sensor. Wu et al. analyzed the decoupling principle of a sliding six-component Force/Torque (F/T) sensor and proposed a robust design method of elastic body size optimization. Experimental results indicate that the nonlinear decoupling methods outperform the traditional linear decoupling methods. The last obtained weight coefficient will determine the BP neural network model which establishes the nonlinear bridge between the input and output of the multi-component force sensor, i.e., the nonlinear static decoupling. As the decoupling of the multi-component force sensor is actually to explore the nonlinear relationship between the multi-channel output voltage signal and the applied multi-component load, the Support Vector Regression algorithm can be adopted to decoupling of multi-component F/T sensing system.
