Search papers, labs, and topics across Lattice.
This paper introduces a novel framework that enhances sign language generation (SLG) outputs for humanoid robots by addressing critical self-intersection issues that arise during execution. By implementing a volumetric collision-mitigation module and a vision-language-guided retargeting algorithm, the authors effectively ensure that generated motions remain physically plausible while adhering closely to the original signing trajectory. The results demonstrate significant improvements in collision handling and perception-guided refinement, which are essential for reliable humanoid signing performance.
Humanoid robots can now execute sign language with reduced collisions and enhanced motion fidelity, thanks to innovative collision-mitigation techniques and vision-language integration.
Recent sign language generation (SLG) systems increasingly output dense 3D body representations, which better preserve full-body kinematics and geometry for downstream embodiment on humanoid robots. However, these generated motions frequently exhibit self-intersections such as hand-hand and hand-torso penetration. While such artifacts may be tolerated in offline rendering, they become critical in humanoid execution as they lead to infeasible inverse-kinematics (IK) solutions, collisions, and unstable retargeted trajectories. We present a system-level framework that bridges SLG outputs to humanoid joint-space execution via two components. First, we introduce a volumetric SMPL-X collision-mitigation module that projects generated signing motions toward physically plausible configurations while minimally deviating from the original trajectory. Second, we propose a vision-language-guided retargeting algorithm built on an IK backbone: a VLM serves as a visual critic over rendered humanoid motion, identifies embodiment-specific failure modes, and triggers targeted task-space corrections. Our results highlight collision handling and perception-guided refinement as key missing components for reliable humanoid signing.
