UniMoGen: Universal Motion Generation

2025-05-28

AI Summary

Existing motion generation methods are constrained by predefined skeletal topologies, limiting generalization across heterogeneous skeletons (e.g., human, animal). This paper introduces the first skeleton-agnostic universal motion generation paradigm: a dynamic graph-based UNet diffusion model. It natively supports variable joint counts, style and trajectory conditioning, and temporal motion continuation—without imposing a fixed maximum joint count—thereby balancing generalizability, controllability, and inference efficiency. The model enables joint training on multi-source heterogeneous datasets (e.g., 100style and LAFAN1). Experiments demonstrate state-of-the-art performance on 100style and robust cross-skeleton generalization: it maintains high fidelity on both LAFAN1 and 100style under joint training, while achieving significantly faster inference compared to prior approaches.

Technology Category

• Computer Vision: Motion & Tracking
• Machine Learning: Deep Generative Models & Autoencoders
• Natural Language Processing: Generation

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Motion generation is a cornerstone of computer graphics, animation, gaming, and robotics, enabling the creation of realistic and varied character movements. A significant limitation of existing methods is their reliance on specific skeletal structures, which restricts their versatility across different characters. To overcome this, we introduce UniMoGen, a novel UNet-based diffusion model designed for skeleton-agnostic motion generation. UniMoGen can be trained on motion data from diverse characters, such as humans and animals, without the need for a predefined maximum number of joints. By dynamically processing only the necessary joints for each character, our model achieves both skeleton agnosticism and computational efficiency. Key features of UniMoGen include controllability via style and trajectory inputs, and the ability to continue motions from past frames. We demonstrate UniMoGen's effectiveness on the 100style dataset, where it outperforms state-of-the-art methods in diverse character motion generation. Furthermore, when trained on both the 100style and LAFAN1 datasets, which use different skeletons, UniMoGen achieves high performance and improved efficiency across both skeletons. These results highlight UniMoGen's potential to advance motion generation by providing a flexible, efficient, and controllable solution for a wide range of character animations.

Problem

• Overcoming skeletal structure limitations in motion generation
• Enabling universal motion generation for diverse characters
• Achieving skeleton-agnostic and computationally efficient motion synthesis

Innovation

• UNet-based diffusion model for motion generation
• Dynamically processes joints for skeleton agnosticism
• Controllable via style and trajectory inputs