Miniature humanoids can perform complex tele-loco-manipulation (remote control of both movement and arm tasks) by combining VR teleoperation for upper-body control with RL-based locomotion—making advanced robot control accessible beyond expensive full-sized platforms.
This paper develops a control system for miniature humanoid robots that combines virtual reality teleoperation for arm control with reinforcement learning for walking and balance. The system was tested on a small ROBOTIS OP3 robot, demonstrating it could walk while manipulating objects and relocate cubes under human remote control.