RETRACTED ARTICLE: Estimation of contact forces of underactuated robotic finger using soft computing methods

Abstract

Underactuated robotic finger could be used as adaptive mechanism with simple control algorithm. In this study the main aim was to estimate the robotic finger contact forces by soft computing methods. Soft computing approach was applied in order to overcome high nonlinearity in the finger behavior. Kinetostatic analysis was performed in order to extract the input/output data samples for the soft computing methods. The main goal was to estimate the contact forces based on contact locations with the objects. Seven soft computing methods were applied: genetic programming, support vector machine, support vector machine with firefly algorithm, artificial neural network, support vector machine with wavelet transfer function), extreme learning machine and extreme learning machine with wavelet transfer function. The reliability of these computational models was analyzed based on simulation results. Extreme learning machine with wavelet transfer function shown the best accuracy for the contact forces estimation.