通过运动具身实现具备物理合理性感知的轨迹预测
Physical Plausibility-aware Trajectory Prediction via Locomotion Embodiment
AI总结：
针对现有人类轨迹预测方法隐式用姿态线索导致结果不合理的问题，提出Locomotion Embodiment框架，通过可微运动价值函数和具身运动损失训练模型，推理时用运动价值滤波器优化，可提升多种HTP方法性能。
摘要：
人类即使仅通过瞬时观察，也能利用与人体姿态相关的线索预测人类未来的轨迹。然而，以往的人类轨迹预测（Human Trajectory Prediction, HTP）方法仅隐式利用姿态线索，导致预测结果缺乏合理性。为解决这一问题，我们提出Locomotion Embodiment（运动具身）框架，该框架通过遵循物理定律的运动生成过程，显式评估预测轨迹的物理合理性。虽然运动合理性是通过不可微的物理模拟器学习得到的，但我们用可微的Locomotion Value（运动价值）函数替代它，以数据驱动的方式训练HTP网络。特别地，我们提出的Embodied Locomotion loss（具身运动损失）有助于利用多头结构高效训练随机HTP网络。此外，我们还提出Locomotion Value filter（运动价值滤波器），用于在推理阶段过滤掉不合理的轨迹。实验表明，我们的方法在多种数据集和问题设置下，甚至能提升当前最优HTP方法的性能。我们的代码可在https://github.com/ImIntheMiddle/EmLoco获取。
英文摘要：
Humans can predict future human trajectories even from momentary observations by using human pose-related cues. However, previous Human Trajectory Prediction (HTP) methods leverage the pose cues implicitly, resulting in implausible predictions. To address this, we propose Locomotion Embodiment, a framework that explicitly evaluates the physical plausibility of the predicted trajectory by locomotion generation under the laws of physics. While the plausibility of locomotion is learned with an indifferentiable physics simulator, it is replaced by our differentiable Locomotion Value function to train an HTP network in a data-driven manner. In particular, our proposed Embodied Locomotion loss is beneficial for efficiently training a stochastic HTP network using multiple heads. Furthermore, the Locomotion Value filter is proposed to filter out implausible trajectories at inference. Experiments demonstrate that our method enhances even the state-of-the-art HTP methods across diverse datasets and problem settings. Our code is available at: https://github.com/ImIntheMiddle/EmLoco.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
