LiveHPS is a novel single-LiDAR-based approach for scene-level 3D human pose and shape estimation in large-scale, uncontrolled environments. The method addresses the challenges of varying light conditions, occlusions, and noise by designing an Adaptive Vertex-guided Distillation module to align diverse point distributions with regular SMPL mesh vertex distributions. It also employs a Consecutive Pose Optimizer to refine joint positions using geometric and dynamic information from consecutive frames, and a Skeleton-aware Translation Solver to estimate global translations accurately. The approach achieves state-of-the-art performance on various datasets, including FreeMotion, a large-scale motion dataset with multi-modal and multi-view data from LiDARs, cameras, and IMUs. Extensive experiments demonstrate the robustness and effectiveness of LiveHPS in real-world applications.LiveHPS is a novel single-LiDAR-based approach for scene-level 3D human pose and shape estimation in large-scale, uncontrolled environments. The method addresses the challenges of varying light conditions, occlusions, and noise by designing an Adaptive Vertex-guided Distillation module to align diverse point distributions with regular SMPL mesh vertex distributions. It also employs a Consecutive Pose Optimizer to refine joint positions using geometric and dynamic information from consecutive frames, and a Skeleton-aware Translation Solver to estimate global translations accurately. The approach achieves state-of-the-art performance on various datasets, including FreeMotion, a large-scale motion dataset with multi-modal and multi-view data from LiDARs, cameras, and IMUs. Extensive experiments demonstrate the robustness and effectiveness of LiveHPS in real-world applications.