Keita Yoneda

Reconfigurable robots address diverse tasks, including locomotion and object manipulation, by changing body configurations. In particular, modular robots expand body functions by connecting multiple body elements into various structures. However, these elements are typically mechanically fixed, and the focus is on functions as a single body after docking. This study proposes TwinX, a robot system in which a single interbody wire connects integrated morphological reconfiguration in the docked state and functional relations among distributed bodies in the separated state. TwinX consists of two wheeled‐legged robots connected by this wire and docks and reconfigures by winding it. Even when the two bodies are distributed, the same wire can be used for object manipulation and force transmission. Real‐robot experiments show quadrupedal and swerve‐drive locomotion in the docked state, as well as object manipulation using the mobile manipulator configuration. They also show enclosure, hooking, and force‐transmission tasks in the distributed state. These results show that the interbody wire extends functional expansion through changes in body configuration to include distributed body relations. Furthermore, it positions the interbody wire as a design element for robot bodies that is different from rigid links and joints because it enables maintained, variable‐length, and tension‐based relations between body elements.

Learning dynamic whole-body motions for legged robots through reinforcement learning (RL) remains challenging due to the high risk of failure, which makes efficient exploration difficult and often leads to unstable learning. In this paper, we propose External Force Guided Curriculum Learning (EFGCL), a guided RL approach based on the principle of physical guidance, in which external assistive forces are introduced during training. Inspired by spotting in artistic gymnastics, EFGCL enables agents to physically experience successful motion executions without relying on task-specific reward shaping or reference trajectories. Experiments on a quadrupedal robot performing Jump, Backflip, and Lateral-Flip tasks demonstrate that EFGCL accelerates learning of the Jump task by approximately a factor of two and enables the acquisition of complex whole-body motions that conventional RL methods fail to learn. We further show that the learned policies can be deployed on a real robot, reproducing motions consistent with those observed in simulation. These results indicate that physically guided exploration, which allows agents to experience success early in training, is an effective and general strategy for improving learning efficiency in dynamic whole-body motion tasks.