Verification of throwing operation by a manipulator with variable viscoelastic joints with straight-fiber-type artificial muscles and magnetorheological brakes

The performance of a robot can be enhanced by increasing its output. However, increasing the output of rigid actuators such as motors and hydraulic actuators will likely increase the weight of the robot. Conversely, organisms such as human beings achieve high output within a short time by accumulating and releasing the elastic energy stored in their muscles; thus, providing an instantaneous force. Moreover, the viscoelastic properties of muscle enable organisms to control their instantaneous force outputs and overall movements. Therefore, in this study, we developed a manipulator with two-degree-of-freedom (DOF) variable viscoelastic joints. The manipulator comprised straight-fiber-type artificial muscles and a magnetorheological (MR) brakes. The ability of a manipulator to generate controlled movement from an instantaneous force was tested in a throwing operation. This simple two-DOF variable viscoelastic manipulator with apparent viscosity control by the MR brakes achieved successful throwing motions. However, it was not possible to calculate the experimental parameters from the target operation in previous study. Therefore, we focused on particle swarm optimization (PSO) as a parameter search method. In this paper, we try to optimize parameters of the throwing movement by combining the spring model of a two-DOF variable viscoelastic manipulator with the PSO method.