Abstract
Exploring the locomotion of creatures is a challenging task in bionic robots, and the existing iterative design methods are mainly based on one or two characteristics to optimize robots. Here, we introduce the thinking of system identification theory to bionic robots, bypassing the exploration of the dynamics and reducing the difficulty of design greatly. A one-degree-of-freedom (DOF) six-bar mechanism (Watt I) was designated as the model to be identified, and it was divided into two parts, i.e., a one-DOF four-bar linkage and a three-DOF series arm. Then, we formed constraints and a loss function. The parameters of the model were identified based on the kinematic data of a jumping marmoset, an animal chosen for its unusually high mass-specific power output. As a result, we obtained the desired model. Then, a prototype derived from the model was fabricated, and the experiments verified the effectiveness of the method. Based on the success of our experiments, we believe our method can be applied to emulate other motions as well.