Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
MaskAdapt introduces a two-stage residual learning framework for physics-based character control, first training a mask-invariant base policy with stochastic masking and action regularization, then training a residual policy to adapt specific body parts. The mask-invariant base policy acts as a robust motion prior, enabling stable control even with missing observations. Experiments demonstrate MaskAdapt's ability to perform motion composition and text-driven partial goal tracking, outperforming existing methods in targeted motion adaptation and robustness.
Achieve flexible and robust physics-based character control by learning a mask-invariant motion prior that anticipates adaptation in masked regions.
We present MaskAdapt, a framework for flexible motion adaptation in physics-based humanoid control. The framework follows a two-stage residual learning paradigm. In the first stage, we train a mask-invariant base policy using stochastic body-part masking and a regularization term that enforces consistent action distributions across masking conditions. This yields a robust motion prior that remains stable under missing observations, anticipating later adaptation in those regions. In the second stage, a residual policy is trained atop the frozen base controller to modify only the targeted body parts while preserving the original behaviors elsewhere. We demonstrate the versatility of this design through two applications: (i) motion composition, where varying masks enable multi-part adaptation within a single sequence, and (ii) text-driven partial goal tracking, where designated body parts follow kinematic targets provided by a pre-trained text-conditioned autoregressive motion generator. Through experiments, MaskAdapt demonstrates strong robustness and adaptability, producing diverse behaviors under masked observations and delivering superior targeted motion adaptation compared to prior work.