Curriculum-reinforcement learning on simulation platform of tendon-driven high-degree of freedom underactuated manipulator

A high degree of freedom (DOF) benefits manipulators by presenting various postures when reaching a target. Using tendon-driven system with an underactuated structure can provide flexibility and weight reduction to such manipulators. The design control composite are challenging owing its complicated architecture modeling difficulties. In our previous study, we developed tendon-driven, high-DOF manipulator inspired from ostrich neck referred as the Robostrich arm. This study particularly focused on problems simulation development manipulator. We proposed curriculum-based reinforcement-learning approach. Inspired human learning, progressing simple complex tasks, arm obtain manipulation abilities step-by-step reinforcement learning ranging position tasks practical application tasks. addition, approach was simulate structure. results show that continuously reach targets simultaneously maintain tip at desired orientation while mounted mobile platform in presence perturbation. These achieve flexible ability even if vibrations presented locomotion.