Athena-WBC: Capability-Aligned Policy Experts for Long-Tail Humanoid Whole-Body Control
2026-07-06 • Robotics
Robotics
AI summaryⓘ
The authors studied how to improve humanoid robots' ability to mimic complex motions that are hard to learn. They found that just training more on difficult moves isn't enough because some motions are too demanding for standard training methods. To fix this, they created Athena-WBC, a system that uses different expert controllers specialized for balance and dynamic movements, which makes learning easier and more effective. Their approach led to better performance on tough motions compared to existing methods, using only a few specialized experts.
humanoid motion trackingpolicy expertswhole-body controlreinforcement learningteacher-student pipelineDAgger distillationbalance controldynamic motionsmotion trackingrobotics training
Authors
Yuan Jiang, Ningyuan Zhang, Xicun Yang, Shidi Li, Yuzhi Jiang, Zhiyi Rong, Shuaikang Ma, Chuanzheng Li, Jie Chen
Abstract
Large-scale humanoid motion-tracking controllers are commonly improved by reallocating training effort: difficult motions are sampled more often, isolated into smaller subsets, or assigned to specialized experts. We show that this view is incomplete. In strong whole-body-control baselines, a residual set of feasible training clips remains unsolved even under targeted training, especially for high-dynamic transitions and balance-critical motions. These failures arise not only from insufficient exposure, but from a mismatch between the motion demands and the effective capability induced by the default training recipe. We propose Athena-WBC, a compact teacher-student pipeline with capability-aligned policy experts for long-tail humanoid whole-body control. Dynamic experts use a tracking-focused, constraint-aware objective that removes conservative effort and temporal-control penalties while preserving physical feasibility constraints; balance experts use a gravity curriculum to improve early-training survivability. The resulting privileged teachers are motion-routed for DAgger distillation and then compressed into a single controller with deployable observations followed by RL fine-tuning. Experiments on a full-size humanoid show improved recovery of training-set long-tail motions and better held-out tracking than a strong SONIC-recipe baseline, using only a small number of experts.