Optimization of Throwing Motion by 2-DOF Variable Viscoelastic Joint Manipulator

This paper focuses on control of variable viscoelasticity joint manipulator. Each joint consists of pneumatic rubber artificial muscle and magnetorheological fluid. And the joint can generate instantaneous force by accumulating potential energy in artificial muscle. Using instantaneous force appropriately, robots can perform dynamic motion such as jumping and throwing like a human. These motions are expected to contribute to the efficient transport of objects and improve the robot's mobility. And also, elasticity and viscosity of joints are needed to control appropriately to achieve target task. Therefore, we proposed a method to control variable viscoelasticity of joint. We set throwing motion as a target task. And elasticity and viscosity are decided by simulation. In addition, simulation result is optimized by Particle Swarm Optimization (PSO) algorithm. Finally, we conducted throwing experiment to reproduce the simulation result. As a result, simulation result showed that elasticity and viscosity changed to accelerate end effector. However, experimental result showed deviations from simulation result because of model error.