国家预印本平台

Human-robot collaboration has been studied primarily in dyadic or sequential settings. However, real homes require multiadic collaboration, where multiple humans and robots share a workspace, acting concurrently on interleaved subtasks with tight spatial and temporal coupling. This regime remains underexplored because close-proximity interaction between humans, robots, and objects creates persistent occlusion and rapid state changes, making reliable real-time 3D tracking the central bottleneck. No existing platform provides the real-time, occlusion-robust, room-scale perception needed to make this regime experimentally tractable. We present OmniRobotHome, the first room-scale residential platform that unifies wide-area real-time 3D human and object perception with coordinated multi-robot actuation in a shared world frame. The system instruments a natural home environment with 48 hardware-synchronized RGB cameras for markerless, occlusion-robust tracking of multiple humans and objects, temporally aligned with two Franka arms that act on live scene state. Continuous capture within this consistent frame further supports long-horizon human behavior modeling from accumulated trajectories. The platform makes the multiadic collaboration regime experimentally tractable. We focus on two central problems: safety in shared human-robot environments and human-anticipatory robotic assistance, and show that real-time perception and accumulated behavior memory each yield measurable gains in both.

$ω$~Centauri, the most massive globular cluster in the Milky Way, exhibits a level of stellar population complexity that has long resisted a unified chemical characterisation. We exploit high-resolution near-infrared spectroscopy from the Milky Way Mapper survey (MWM DR19) to construct one of the largest homogeneously analysed samples of $ω$~Cen members to date. Applying Ward-linkage hierarchical clustering in a seven-dimensional chemical abundance space, without prior assumptions on population number or boundaries, we identify ten chemically distinct stellar populations. Their nucleosynthetic signatures trace four enrichment channels: iron-peak, $α$-element, CNO-cycle, and high-temperature proton-capture processes. The populations organise into two dominant groups separated by a large light-element spread at a modest iron baseline, consistent with AGB-driven self-enrichment. This dichotomy reflects distinct enrichment pathways: core-collapse supernovae establish the iron baseline, while AGB stars dominate light-element and $s$-process enrichment. A decoupled rise in $s$-process abundances relative to iron-peak elements, together with sub-dominant Type~Ia contributions across all metallicities, indicates evolution on timescales shorter than the characteristic Type~Ia delay time. One intermediate-metallicity population retains a primordial composition, providing evidence for spatially segregated enrichment within the progenitor. The most metal-rich component may trace star formation continuing after accretion into the Milky Way halo. All populations lie in the accreted regime of the $[\mathrm{Al/Fe}]$--$[\mathrm{Mg/Mn}]$ plane, supporting an ex-situ origin. These results reinforce the interpretation of $ω$~Cen as the remnant nucleus of an accreted dwarf galaxy and provide a framework for future chemo-dynamical studies.

The Karch-Randall braneworld concerns the physics of an AdS$_{d}$ brane embedded in an ambient gravitational AdS$_{d+1}$ spacetime. The gravitational theory induced on the AdS$_{d}$ brane has a very light but massive graviton. It has been established that the zero graviton mass limit of the $d$-dimensional graviton propagator is smooth at tree-level. Furthermore, this smoothness was conjectured to persist to the quantum level. This conjecture suggests that the massive graviton on the AdS$_{d}$ brane is due to spontaneous symmetry breaking, which is consistent with its holographic dual description. In this letter, we show that the zero mass limit of the partition function is a theory of a massless graviton and a decoupled massive vector. The zero mass limit is not the basic Randall-Sundrum II model, but a theory with these additional decoupled vector degrees of freedom coupled only to gravity. The proof relies on deriving the fully covariant description of the $d$-dimensional gravity theory which enables us to compute the one-loop partition function. At the end, we comment on the implications of this result to the physics of entanglement islands.

Reconstructing 3D scenes from sparse, unposed images remains challenging under real-world conditions with varying illumination and transient occlusions. Existing methods rely on scene-specific optimization using appearance embeddings or dynamic masks, which requires extensive per-scene training and fails under sparse views. Moreover, evaluations on limited scenes raise questions about generalization. We present GenWildSplat, a feed-forward framework for sparse-view outdoor reconstruction that requires no per-scene optimization. Given unposed internet images, GenWildSplat predicts depth, camera parameters, and 3D Gaussians in a canonical space using learned geometric priors. An appearance adapter modulates appearance for target lighting conditions, while semantic segmentation handles transient objects. Through curriculum learning on synthetic and real data, GenWildSplat generalizes across diverse illumination and occlusion patterns. Evaluations on PhotoTourism and MegaScenes benchmark demonstrate state-of-the-art feed-forward rendering quality, achieving real-time inference without test-time optimization

Vision-Language-Action (VLA) models have increasingly incorporated reasoning mechanisms for complex robotic manipulation. However, existing approaches share a critical limitation: whether employing explicit linguistic reasoning that suffers from latency and discretization, or utilizing more expressive continuous latent reasoning, they are predominantly confined to static imitation learning that limits adaptability and generalization. While online reinforcement learning (RL) has been introduced to VLAs to enable trial-and-error exploration, current methods exclusively optimize the vanilla action space, bypassing the underlying physical reasoning process. In this paper, we present \textbf{LaST-R1}, a unified VLA framework that integrates latent Chain-of-Thought (CoT) reasoning over physical dynamics prior to action execution, along with a tailored RL post-training paradigm. Specifically, we propose \textbf{Latent-to-Action Policy Optimization (LAPO)}, a novel RL algorithm that jointly optimizes the latent reasoning process and the action generation. By bridging reasoning and control, LAPO improves the representation of physical world modeling and enhances robustness in interactive environments. Furthermore, an \textbf{adaptive latent CoT mechanism} is introduced to allow the policy to dynamically adjust its reasoning horizon based on environment complexity. Extensive experiments show that LaST-R1 achieves a near-perfect 99.8\% average success rate on the LIBERO benchmark with only one-shot supervised warm-up, significantly improving convergence speed and performance over prior state-of-the-art methods. In real-world deployments, LAPO post-training yields up to a 44\% improvement over the initial warm-up policy across four complex tasks, including both single-arm and dual-arm settings. Finally, LaST-R1 demonstrates strong generalization across simulated and real-world environments.