用单阶段扩散模型提升人体运动预测的时空一致性与多样性。
SMamDiff: Spatial Mamba for Stochastic Human Motion Prediction
- 通过残差DCT编码消除位置偏移，聚焦关节运动模式。
- 采用顺序处理的骨架注意力模块，建模长程跨关节依赖。
- 在主流数据集上性能领先，且更适合边缘部署。
随着智能室内感知和服务机器人广泛应用，人体运动预测（HMP）对实现安全、主动的辅助至关重要。然而，现有方法要么生成单一确定性预测而忽略不确定性，要么依赖概率模型却牺牲了运动合理性。扩散模型虽提升了准确率与多样性权衡，但通常依赖多阶段流程，难以在边缘设备部署。本文聚焦于如何在单阶段扩散模型中保持时空一致性，提出SMamDiff：一种基于空间Mamba的扩散模型，包含两项创新设计：(i) 残差-DCT运动编码，在时域DCT前减去最后一个观测姿态，降低直流分量（f=0）主导性，突出高阶频率信息，使模型学习关节运动而非位置；(ii) 基于绘图式顺序处理的骨架-马尔可夫模块，按关节顺序逐个处理，使后序关节依赖先前关节，从而建立长程跨关节依赖关系。在Human3.6M和HumanEva数据集上，该方法在单阶段概率型HMP中达到最先进性能，同时相比多阶段扩散基线显著降低延迟与内存占用。
原文摘要 · Abstract (English)
With intelligent room-side sensing and service robots widely deployed, human motion prediction (HMP) is essential for safe, proactive assistance. However, many existing HMP methods either produce a single, deterministic forecast that ignores uncertainty or rely on probabilistic models that sacrifice kinematic plausibility. Diffusion models improve the accuracy-diversity trade-off but often depend on multi-stage pipelines that are costly for edge deployment. This work focuses on how to ensure spatial-temporal coherence within a single-stage diffusion model for HMP. We introduce SMamDiff, a Spatial Mamba-based Diffusion model with two novel designs: (i) a residual-DCT motion encoding that subtracts the last observed pose before a temporal DCT, reducing the first DC component ($f=0$) dominance and highlighting informative higher-frequency cues so the model learns how joints move rather than where they are; and (ii) a stickman-drawing spatial-mamba module that processes joints in an ordered, joint-by-joint manner, making later joints condition on earlier ones to induce long-range, cross-joint dependencies. On Human3.6M and HumanEva, these coherence mechanisms deliver state-of-the-art results among single-stage probabilistic HMP methods while using less latency and memory than multi-stage diffusion baselines.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。