Identification design of one-DOF eight-bar based on limb movements

Abstract

This paper presents an identification design of one-DOF eight-bar based on limb movements by splitting it into a Stephenson II six-bar linkage and a 3R serial chain. They are kinematically identified, followed motion coupling, to form a desired eight-bar model. Specifically, enlightened by the system identification theory which achieves the behavioral simulation instead of focusing on their principles, we extend this idea to the mechanism design field to realize a bionic design. The steps are as follows: (1) Extracting input/output data from a movement, (2) forming a model set with appropriate capacities, (3) defining a criterion as the identification goal, (4) implementing an identification. Two instances involved in the identification of the movements of human fingers and marmoset legs are implemented to show the identification procedures in detail. Then we apply the genetic algorithms (GA) to obtain the expectation of eight-bar models. Ascribe to the identification method, the two models will be developed as a finger grasping device and a jumping robot respectively, achieving some principles that are ignored at the beginning.