EFGCL: Learning Dynamic Motion through Spotting-Inspired External Force Guided Curriculum Learning
2026-05-11 • Robotics
Robotics
AI summaryⓘ
The authors address the difficulty of teaching legged robots to do complex moves using reinforcement learning, which often struggles because the robots fail too much early on. They introduce a method called External Force Guided Curriculum Learning (EFGCL), where the robot gets physical help during training, similar to how a gymnast is assisted while learning new moves. This helps the robot experience success sooner without needing special rewards or example motions. Their tests with a four-legged robot show that EFGCL speeds up learning and enables moves that normal methods can't achieve, and the learned moves work well on a real robot too.
reinforcement learningcurriculum learninglegged robotsdynamic whole-body motionexternal force guidancephysical guidancerobot locomotionrobot learningsimulation to real transferrobotics
Authors
Keita Yoneda, Kento Kawaharazuka, Kei Okada
Abstract
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 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.