AtomicMotion: Learning Human Motion From Different Human Parts

2026-05-21

AI Summary

This work addresses the limitations of existing whole-body pose reconstruction methods, which treat the human body as a monolithic entity and often suffer from error accumulation and incoherent joint motions. To overcome these issues, the authors propose a functionally motivated five-cluster logical body partitioning strategy, integrated with mask-based full-body pre-conditioning training and a Kinematic Attention mechanism that incorporates skeletal kinematic tree structures. This approach effectively decouples and coordinates the motion dynamics of individual body parts. Evaluated on the AMASS dataset, the method significantly outperforms current state-of-the-art techniques, achieving higher reconstruction accuracy under sparse head and hand trajectory inputs while also enhancing biomechanical plausibility and fine-grained motion expressiveness.

Abstract

Accurately reconstructing full-body poses from sparse head and hand trajectories is a foundational challenge for immersive AR/VR telepresence. Current methods often struggle with error accumulation and unnatural joint coordination, primarily because they treat the human body as a monolithic entity, thereby failing to capture the fine-grained ``atomic intents'' embedded in subtle signal variations and overlooking the inherent structural topology. To bridge this gap, we present AtomicMotion, a framework designed to decouple and re-integrate body dynamics through three core innovations. First, we introduce a logical body partitioning scheme that decomposes the skeleton into five distinct clusters based on functional intent; this ensures that each partition preserves internal joint synergies while isolating local motion primitives. Second, to robustly map sparse inputs to high-dimensional poses, we employ a masked full-body pre-conditioning strategy during training, forcing the model to internalize global skeletal topology and latent kinematic constraints. Finally, addressing the limitations of vanilla spatial attention, which often ignores fixed physiological connectivity, we propose Kinematic Attention. By embedding the classical kinematic tree structure into the attention mechanism, we ensure biological plausibility in the synthesized motions. Extensive evaluations on the AMASS dataset demonstrate that AtomicMotion significantly outperforms existing baselines, yielding higher reconstruction fidelity and superior biomechanical realism.

Problem

• full-body pose reconstruction
• sparse motion input
• human motion modeling
• AR/VR telepresence
• biomechanical realism

Innovation

• AtomicMotion
• logical body partitioning
• masked full-body pre-conditioning
• Kinematic Attention
• motion reconstruction