A MLP-Hedge-Algebras Admittance Controller for Physical Human–Robot Interaction

Recently, the identification of inertia and damping matrices (IIDM) safety issues, as well natural cooperation, are interestingly considered to enhance quality physical human–robot interaction (pHRI). To cover all these advanced admittance controllers, such those based on fuzzy logic or hedge algebras, have been formulated successfully applied in several industrial problems. However, inference mechanism kinds controllers causes discreteness super surface describing input–output relationship Cartesian coordinates. As a consequence, safe-natural cooperation between humans robots is negatively affected. This paper presents an alternative controller for pHRI by using combination algebras multilayer perceptron neural network (MLP), whose purpose create more accurate controller. our best knowledge, this first time that also used pHRI. The proposed verified teaching task 6-DOF manipulator. Experimental results shown method provides better compared with previous methods.