GST-Bench: Can VLMs Develop Global Spatial Awareness from Video?

Spatial intelligence is fundamental to embodied agents, yet existing benchmarks focus on local spatial perception from single or few viewpoints, overlooking global spatial awareness over continuous, long-horizon visual streams. To address this limitation, we introduce the Global-Spatial-Temporal Benchmark (GST-Bench), a VQA benchmark for global spatial intelligence in video understanding, comprising human-verified questions derived from 6,790 minutes of synthetically generated video. It requires models to perform accurate spatial inference from novel viewpoints unseen in the input video and to map egocentric observations onto global top-down images. A comprehensive evaluation of 22 state-of-the-art VLMs exposes a striking gap between models and humans: the strongest zero-shot model attains only 42.68, far below the human score of 79.08. To probe the cause of this gap, we construct GST-Bench-Local and find that models, despite strong local spatial understanding under the same task formulation, still fail to consolidate long-horizon observations into a globally consistent scene representation. We further provide GST-Train, a dataset for global spatial reasoning, as a complementary resource to facilitate future research on this challenge.

Bayesian Bandit Beamforming with Implicit Channel Learning for RIS under Hybrid Near/Far-Field Propagation

Reconfigurable intelligent surfaces (RISs) can improve high-frequency wireless links by shaping the propagation environment, but their passive architecture makes channel acquisition costly. Conventional estimate-then-optimize methods usually require pilot overhead that scales with the number of reflecting elements, which is undesirable under short coherence times and hybrid near-/far-field propagation. This paper proposes the Bayesian bandit framework for RIS phase-shift configuration with implicit channel learning. The method updates a Gaussian posterior of the cascaded channel from one scalar pilot observation per slot and uses Thompson sampling to balance channel estimation and beamforming gain. We derive a Bayesian regret decomposition that connects Bayesian received-power regret to posterior uncertainty contraction, and further establish a conditional sublinear Bayesian-regret guarantee. To exploit sparse hybrid-field propagation, we develop an energy-focusing angle-distance dictionary and a sparse Bayesian learning (SBL)-based Thompson-sampling algorithm with warm-started hyperparameter refinement. Simulations show that the proposed policies approach the perfect-channel state information (CSI) benchmark in the line-of-sight (LOS)-dominant setting within 10 time block and improve transmission efficiency over the considered baselines in multipath and Rayleigh fading scenarios.

Exploratory, Communicative, and Deployable: Vision-Driven Embodied Agents for Open-World Mobile Manipulation

Real-world deployment of embodied agents requires active exploration, visual grounding, and interactive intent disambiguation. However, existing frameworks often rely on privileged simulator states or assume complete instructions, bypassing realistic deployment challenges. To bridge this gap, we present REAL, an agentic framework for open-world mobile manipulation. REAL establishes sim-to-real-consistent environment APIs without oracle perception and integrates a simulated user to enable human-in-the-loop interaction. Within this environment, we design diverse task compositions to drive data collection, supervised fine-tuning, and online reinforcement learning, systematically optimizing agent performance. To comprehensively evaluate this approach, we introduce REAL-Bench, a benchmark spanning 241 tasks across active exploration, visual distraction, articulated manipulation, and interactive disambiguation. Experimental results demonstrate that our trained agent outperforms leading commercial closed-source VLMs on interactive tasks with a 56.9% success rate. Further empirical analysis reveals that our hierarchical training pipeline successfully aligns the model's tool-use capabilities while maintaining robust open-vocabulary reasoning under extended exploration horizons. Finally, we deploy and evaluate our framework on a physical dual-arm mobile robot, where it achieves a 78.3% end-to-end success rate over 60 real-world episodes. These physical trials demonstrate robust zero-shot transferability to unseen household scenarios, validating that our sim-to-real-consistent design successfully bridges the reality gap for long-horizon mobile manipulation. Code is available at https://github.com/InternRobotics/REAL.

Bounded Deep Unfolding for Joint Beamforming and Scheduling in Multi-Cell MIMO Networks

This paper investigates the joint resource block group (RBG) scheduling and beamforming optimization problem for weighted sum-rate (WSR) maximization in multi-cell multiple-input multiple-output (MIMO) downlink networks. While the Fast Fractional Programming (FastFP) framework provides a reliable model-driven solution, it suffers from conservative continuous beamforming updates and prohibitive computational overhead during the discrete RBG matching phase. To address these bottlenecks, we propose a joint deep unfolding framework comprising two core modules: P-Net and K-Net. For continuous beamforming, P-Net learns an adaptive relaxation factor along the analytical FastFP update direction. By strictly constraining this factor within an ascent-preserving interval, P-Net accelerates the optimization trajectory while rigorously retaining monotonic improvement and stationary-point convergence guarantees. For discrete RBG scheduling, K-Net learns a long-horizon priority policy that guides a low-complexity greedy assignment, effectively preserving the assignment quality while bypassing the high complexity of Hungarian matching. Both networks leverage analytical algorithmic priors and utilize recurrent parameter sharing, enabling flexible inference beyond the training horizon. Extensive simulations demonstrate that the proposed joint framework achieves higher WSR and faster execution times than conventional model-driven baselines, while generalizing robustly across unseen network scales, antenna configurations, and channel conditions without retraining.

Environment-Conditioned Diffusion Meta-Learning for Data-Efficient WiFi Localization

Fingerprinting-based localization often suffers from poor cross-environment generalization, especially when only a few labeled samples are available in the target environment. Existing methods mitigate distribution shifts through domain adaptation or improved signal representations, but they usually ignore environmental geometry or use it in a deterministic manner, limiting their ability to capture diverse multipath variations in complex propagation conditions. To address this issue, we propose EnvCoLoc, an environment-conditioned diffusion meta-learning framework for few-shot fingerprinting localization. EnvCoLoc extracts structured descriptors from 3D point clouds and uses them to condition a latent diffusion generator, which produces environment-specific parameter offsets to modulate a shared meta-learned initialization. This design injects geometry-aware priors into the adaptation process and provides more informative initializations for new environments. To learn the stochastic mapping from coarse environmental descriptors to high-dimensional parameter corrections under limited data, the diffusion generator and localization network are jointly optimized within a two-loop meta-learning framework. The generated offsets capture systematic environment-dependent variations, while gradient-based inner-loop adaptation further refines the model to reduce residual task-specific mismatch. We also provide an excess-loss analysis for finite-step adaptation, theoretically supporting the benefit of geometry-aware initialization. Real-world experiments show that EnvCoLoc consistently improves localization accuracy over baseline methods, achieving up to a 20.0% reduction in mean localization error in NLOS scenarios with only 10 support samples.

PILOT: One Physics-Integrated Generation Framework to Unify 2D and 3D Radio Map Construction

Unified 2D and 3D radio map construction supports network planning, wireless digital twins, and unmanned aerial vehicle (UAV) applications. In urban environments, blockage, reflection, and diffraction make accurate construction expensive for physics-based solvers. Autoregressive next-token prediction offers a single sequential formulation that can cover both 2D and 3D generation, but standard raster ordering ignores the spatial structure of radio propagation. When generation follows propagation, each token is predicted from propagation-relevant history rather than spatially arbitrary context, which provides more causally informative conditioning and lowers conditional uncertainty. We propose PILOT, a pretrained autoregressive framework that replaces raster scan with a wavefront sequence expanding outward from the transmitter. Each prediction step is guided by an environment-aware instruction that spatially aligns environment features with the queried radio map region. The same framework extends to 3D radio maps through height-slice stacking while a gradient loss enforces vertical continuity. On standard 2D benchmarks, PILOT achieves the lowest NMSE among all baselines. For volumetric generation, it reduces NMSE by 78% relative to the diffusion baseline at roughly $2500\times$ faster inference. It also outperforms methods that rely on 10% sparse measurements and achieves the best zero-shot results in the cross-domain evaluation.

Robust Semantic Transmission for Low-Altitude UAVs: Predictive Channel-Aware Scheduling and Generative Reconstruction

Unmanned aerial vehicle (UAV) downlink transmission facilitates critical time-sensitive visual applications but is fundamentally constrained by bandwidth scarcity and dynamic channel impairments. The rapid fluctuation of the air-to-ground (A2G) link creates a regime where reliable transmission slots are intermittent and future channel quality can only be predicted with uncertainty. Conventional deep joint source-channel coding (DeepJSCC) methods transmit coupled feature streams, causing global reconstruction failure when specific time slots experience deep fading. Decoupling semantic content into a deterministic structure component and a stochastic texture component enables differentiated error protection strategies aligned with channel reliability. A predictive transmission framework is developed that utilizes a split-stream variational codec and a channel-aware scheduler to prioritize the delivery of structural layout over reliable slots. Experimental evaluations indicate that this approach achieves a 5.6 dB gain in peak signal-to-noise (SNR) ratio over single-stream baselines and maintains structural fidelity under significant prediction mismatch.

Physics-Aware Tensor Reconstruction for Radio Maps in Pixel-Based Fluid Antenna Systems

The deployment of pixel-based antennas and fluid antenna systems (FAS) is hindered by prohibitive channel state information (CSI) acquisition overhead. While radio maps enable proactive mode selection, reconstructing high-fidelity maps from sparse measurements is challenging. Existing physics-agnostic or data-driven methods often fail to recover fine-grained shadowing details under extreme sparsity. We propose a Physics-Regularized Low-Rank Tensor Completion (PR-LRTC) framework for radio map reconstruction. By modeling the signal field as a three-way tensor, we integrate environmental low-rankness with deterministic antenna physics. Specifically, we leverage Effective Aerial Degrees-of-Freedom (EADoF) theory to derive a differential gain topology map as a physical prior for regularization. The resulting optimization problem is solved via an efficient Alternating Direction Method of Multipliers (ADMM)-based algorithm. Simulations show that PR-LRTC achieves a 4 dB gain over baselines at a 10% sampling ratio. It effectively preserves sharp shadowing edges, providing a robust, physics-compliant solution for low-overhead beam management.

MIMO Beam Map Reconstruction via Toeplitz-Structured Matrix-Vector Tensor Decomposition

As wireless networks progress toward sixthgeneration (6G), understanding the spatial distribution of directional beam coverage becomes increasingly important for beam management and link optimization. Multiple-input multipleoutput (MIMO) beam map provides such spatial awareness, yet accurate construction under sparse measurements remains difficult due to incomplete spatial coverage and strong angular variations. This paper presents a tensor decomposition approach for reconstructing MIMO beam map from limited measurements. By transforming measurements from a Cartesian coordinate system into a polar coordinate system, we uncover a matrix-vector outer-product structure associated with different propagation conditions. Specifically, we mathematically demonstrate that the matrix factor, representing beam-space gain, exhibits an intrinsic Toeplitz structure due to the shift-invariant nature of array responses, and the vector factor captures distance-dependent attenuation. Leveraging these structural priors, we formulate a regularized tensor decomposition problem to jointly reconstruct line-of-sight (LOS), reflection, and obstruction propagation conditions. Simulation results confirm that the proposed method significantly enhances data efficiency, achieving a normalized mean square error (NMSE) reduction of over 20% compared to state-of-the-art baselines, even under sparse sampling regimes.

LISN: Language-Instructed Social Navigation with VLM-based Controller Modulating

Towards human-robot coexistence, socially aware navigation is significant for mobile robots. Yet existing studies on this area focus mainly on path efficiency and pedestrian collision avoidance, which are essential but represent only a fraction of social navigation. Beyond these basics, robots must also comply with user instructions, aligning their actions to task goals and social norms expressed by humans. In this work, we present LISN-Bench, the first simulation-based benchmark for language-instructed social navigation. Built on Rosnav-Arena 3.0, it is the first standardized social navigation benchmark to incorporate instruction following and scene understanding across diverse contexts. To address this task, we further propose Social-Nav-Modulator, a fast-slow hierarchical system where a VLM agent modulates costmaps and controller parameters. Decoupling low-level action generation from the slower VLM loop reduces reliance on high-frequency VLM inference while improving dynamic avoidance and perception adaptability. Our method achieves an average success rate of 91.3%, which is greater than 63% than the most competitive baseline, with most of the improvements observed in challenging tasks such as following a person in a crowd and navigating while strictly avoiding instruction-forbidden regions. The project website is at: https://social-nav.github.io/LISN-project/