面向物体引导的人-人协同操作的稳定性驱动运动生成
Stability-Driven Motion Generation for Object-Guided Human-Human Co-Manipulation
把流匹配的向量场回归与基于采样的稳定性优化耦合起来，用于生成多个人体与共享物体之间的接触运动；其中「对抗交互先验」与「仿真驱动的流场修正」这两个模块，可以直接迁移到人-物交互、双手操作等需要物理合理性的运动生成任务上。
原文依据 we also incorporate a stability-driven simulation into the flow matching process, which refines unstable interaction states through sampling-based optimization
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论协同操作要求多个人在共享同一物体的同时同步各自的动作，并保证交互合理、姿态自然、状态稳定。然而，现有的大多数运动生成方法面向单角色场景，或者未能考虑负载引发的动力学。本文提出一个流匹配框架，使生成的协同操作运动既符合预期目标，又保持自然性与有效性。具体而言，我们首先引入一个生成模型，从物体的可供性与空间构型中推导出显式的操作策略，用以引导运动流走向成功的操作。为提升运动质量，我们随后设计了一个对抗交互先验，在协同操作过程中促进个体姿态的自然性与人际交互的真实性。此外，我们还将稳定性驱动的仿真融入流匹配过程，通过基于采样的优化修正不稳定的交互状态，并直接调整向量场回归，以促进更有效的操作。实验结果表明，与最优的人-物交互基线相比，本方法取得了更高的接触精度、更低的穿透以及更好的分布保真度。代码将公开。
英文原文摘要 1415 字符
Co-manipulation requires multiple humans to synchronize their motions with a shared object while ensuring reasonable interactions, maintaining natural poses, and preserving stable states. However, most existing motion generation approaches are designed for single-character scenarios or fail to account for payload-induced dynamics. In this work, we propose a flow-matching framework that ensures the generated co-manipulation motions align with the intended goals while maintaining naturalness and effectiveness. Specifically, we first introduce a generative model that derives explicit manipulation strategies from the object's affordance and spatial configuration, which guide the motion flow toward successful manipulation. To improve motion quality, we then design an adversarial interaction prior that promotes natural individual poses and realistic inter-person interactions during co-manipulation. In addition, we also incorporate a stability-driven simulation into the flow matching process, which refines unstable interaction states through sampling-based optimization and directly adjusts the vector field regression to promote more effective manipulation. The experimental results demonstrate that our method achieves higher contact accuracy, lower penetration, and better distributional fidelity compared to state-of-the-art human-object interaction baselines. The code will be made publicly available.