Causal Motion Diffusion Models for Autoregressive Motion Generation

TL;DR

This paper introduces Causal Motion Diffusion Models (CMDM), a novel autoregressive framework that enables real-time, high-quality human motion synthesis by combining causal diffusion transformers with semantic latent representations.

Contribution

The work presents a unified causal diffusion transformer framework for autoregressive motion generation, improving temporal causality, inference speed, and motion quality over prior diffusion and autoregressive models.

Findings

Outperforms existing models in semantic fidelity and smoothness.

Reduces inference latency significantly.

Supports real-time, long-horizon motion generation.

Abstract

Recent advances in motion diffusion models have substantially improved the realism of human motion synthesis. However, existing approaches either rely on full-sequence diffusion models with bidirectional generation, which limits temporal causality and real-time applicability, or autoregressive models that suffer from instability and cumulative errors. In this work, we present Causal Motion Diffusion Models (CMDM), a unified framework for autoregressive motion generation based on a causal diffusion transformer that operates in a semantically aligned latent space. CMDM builds upon a Motion-Language-Aligned Causal VAE (MAC-VAE), which encodes motion sequences into temporally causal latent representations. On top of this latent representation, an autoregressive diffusion transformer is trained using causal diffusion forcing to perform temporally ordered denoising across motion frames. To achieve fast inference, we introduce a frame-wise sampling schedule with causal uncertainty, where each subsequent frame is predicted from partially denoised previous frames. The resulting framework supports high-quality text-to-motion generation, streaming synthesis, and long-horizon motion generation at interactive rates. Experiments on HumanML3D and SnapMoGen demonstrate that CMDM outperforms existing diffusion and autoregressive models in both semantic fidelity and temporal smoothness, while substantially reducing inference latency.