Track new results across pure and applied mathematics.
Search papers, labs, and topics across Lattice.
To bridge high-level robotic foundation planners and low-level humanoid controllers, this work formulates "spectral skills," a compact latent interface trained via predictive representation learning rather than autoencoding reconstruction. By structuring the latent space to forecast subsequent dynamics, the learned interface enables linear vector-space composition of novel behaviors and zero-shot skill chaining without specialized transition policies. On a 29-DoF Unitree G1 humanoid, a frozen controller conditioned on spectral skills cuts global tracking error by 62% over prior state-of-the-art and successfully grounds language-planned composite behaviors in hardware.
Predicting future motion instead of reconstructing past trajectories allows a frozen humanoid controller to cut tracking error by 62% while unlocking zero-shot behavior composition via simple latent addition.
Robotic foundation models offer a promising path toward general-purpose humanoid robot control, often through hierarchical architectures. However, their effectiveness depends on the command interface between the planner and the controller, which must support accurate execution while remaining easy to predict, and ideally allow new behaviors to be composed from prior ones. In this work, we introduce spectral skills, a latent representation of this interface that meets these requirements through predictive representation learning. By design, spectral skills compactly encode short motion segments and are learned by predicting subsequent motion rather than reconstructing the encoder input. On a 29-DoF humanoid, a controller conditioned on spectral skills reduces global tracking error by 62\% relative to the state of the art. The same frozen controller chains independently encoded skills without a separate transition policy. It also composes new behaviors by adding orthogonal directions to any compatible base skill, producing combinations unseen in the training data. We demonstrate tracking, chaining, and composition, as well as control through a language-conditioned planner, on Unitree G1 hardware. Project page: https://spectral-skill.github.io