HiCI: Hierarchical Construction-Integration for Long-Context Attention

Long-context language modeling is commonly framed as a scalability challenge of token-level attention, yet local-to-global information structuring remains largely implicit in existing approaches. Drawing on cognitive theories of discourse comprehension, we propose HiCI (Hierarchical Construction--Integration), a hierarchical attention module that constructs segment-level representations, integrates them into a shared global context, and broadcasts both to condition segment-level attention. We validate HiCI through parameter-efficient adaptation of LLaMA-2 with only <5.5% additional parameters, extending context from 4K to 100K tokens (7B) and 64K tokens (13B). Across language modeling, retrieval, and instruction-following benchmarks, HiCI yields consistent improvements over strong baselines, including matching proprietary models on topic retrieval and surpassing GPT-3.5-Turbo-16K on code comprehension. These results demonstrate the effectiveness of explicit hierarchical structuring as an inductive bias for long-context modeling.

SmoothTurn: Learning to Turn Smoothly for Agile Navigation with Quadrupedal Robots

Quadrupedal robots show great potential for valuable real-world applications such as fire rescue and industrial inspection. Such applications often require urgency and the ability to navigate agilely, which in turn demands the capability to change directions smoothly while running in high speed. Existing approaches for agile navigation typically learn a single-goal reaching policy by encouraging the robot to stay at the target position after reaching there. As a result, when the policy is used to reach sequential goals that require changing directions, it cannot anticipate upcoming maneuvers or maintain momentum across the switch of goals, thereby preventing the robot from fully exploiting its agility potential. In this work, we formulate the task as sequential local navigation, extending the single-goal-conditioned local navigation formulation in prior work. We then introduce SmoothTurn, a learning-based control framework that learns to turn smoothly while running rapidly for agile sequential local navigation. The framework adopts a novel sequential goal-reaching reward, an expanded observation space with a lookahead window for future goals, and an automatic goal curriculum that progressively expands the difficulty of sampled goal sequences based on the goal-reaching performance. The trained policy can be directly deployed on real quadrupedal robots with onboard sensors and computation. Both simulation and real-world empirical results show that SmoothTurn learns an agile locomotion policy that performs smooth turning across goals, with emergent behaviors such as controlling momentum when switching goals, facing towards the future goal in advance, and planning efficient paths. We have provided video demos of the learned motions in the supplementary materials. The source code and trained policies will be made available upon acceptance.

Affordance Field Intervention: Enabling VLAs to Escape Memory Traps in Robotic Manipulation

Vision-Language-Action (VLA) models have shown great performance in robotic manipulation by mapping visual observations and language instructions directly to actions. However, they remain brittle under distribution shifts: when test scenarios change, VLAs often reproduce memorized trajectories instead of adapting to the updated scene, which is a failure mode we refer to as the "Memory Trap". This limitation stems from the end-to-end design, which lacks explicit 3D spatial reasoning and prevents reliable identification of actionable regions in unfamiliar environments. To compensate for this missing spatial understanding, 3D Spatial Affordance Fields (SAFs) can provide a geometric representation that highlights where interactions are physically feasible, offering explicit cues about regions the robot should approach or avoid. We therefore introduce Affordance Field Intervention (AFI), a lightweight hybrid framework that uses SAFs as an on-demand plug-in to guide VLA behavior. Our system detects memory traps through proprioception, repositions the robot to recent high-affordance regions, and proposes affordance-driven waypoints that anchor VLA-generated actions. A SAF-based scorer then selects trajectories with the highest cumulative affordance. Extensive experiments demonstrate that our method achieves an average improvement of 23.5% across different VLA backbones ($π_{0}$ and $π_{0.5}$) under out-of-distribution scenarios on real-world robotic platforms, and 20.2% on the LIBERO-Pro benchmark, validating its effectiveness in enhancing VLA robustness to distribution shifts.

Action-aware Dynamic Pruning for Efficient Vision-Language-Action Manipulation

Robotic manipulation with Vision-Language-Action models requires efficient inference over long-horizon multi-modal context, where attention to dense visual tokens dominates computational cost. Existing methods optimize inference speed by reducing visual redundancy within VLA models, but they overlook the varying redundancy across robotic manipulation stages. We observe that the visual token redundancy is higher in coarse manipulation phase than in fine-grained operations, and is strongly correlated with the action dynamic. Motivated by this observation, we propose \textbf{A}ction-aware \textbf{D}ynamic \textbf{P}runing (\textbf{ADP}), a multi-modal pruning framework that integrates text-driven token selection with action-aware trajectory gating. Our method introduces a gating mechanism that conditions the pruning signal on recent action trajectories, using past motion windows to adaptively adjust token retention ratios in accordance with dynamics, thereby balancing computational efficiency and perceptual precision across different manipulation stages. Extensive experiments on the LIBERO suites and diverse real-world scenarios demonstrate that our method significantly reduces FLOPs and action inference latency (\textit{e.g.} $1.35 \times$ speed up on OpenVLA-OFT) while maintaining competitive success rates (\textit{e.g.} 25.8\% improvements with OpenVLA) compared to baselines, thereby providing a simple plug-in path to efficient robot policies that advances the efficiency and performance frontier of robotic manipulation. Our project website is: \href{https://vla-adp.github.io/}{ADP.com}.

Marine Saliency Segmenter: Object-Focused Conditional Diffusion with Region-Level Semantic Knowledge Distillation

Marine Saliency Segmentation (MSS) plays a pivotal role in various vision-based marine exploration tasks. However, existing marine segmentation techniques face the dilemma of object mislocalization and imprecise boundaries due to the complex underwater environment. Meanwhile, despite the impressive performance of diffusion models in visual segmentation, there remains potential to further leverage contextual semantics to enhance feature learning of region-level salient objects, thereby improving segmentation outcomes. Building on this insight, we propose DiffMSS, a novel marine saliency segmenter based on the diffusion model, which utilizes semantic knowledge distillation to guide the segmentation of marine salient objects. Specifically, we design a region-word similarity matching mechanism to identify salient terms at the word level from the text descriptions. These high-level semantic features guide the conditional feature learning network in generating salient and accurate diffusion conditions with semantic knowledge distillation. To further refine the segmentation of fine-grained structures in unique marine organisms, we develop the dedicated consensus deterministic sampling to suppress overconfident missegmentations. Comprehensive experiments demonstrate the superior performance of DiffMSS over state-of-the-art methods in both quantitative and qualitative evaluations.

VLA-Cache: Towards Efficient Vision-Language-Action Model via Adaptive Token Caching in Robotic Manipulation

Vision-Language-Action (VLA) model can process instructions and visual perception to directly generate actions as output in an end-to-end fashion due to its strong multi-modal reasoning capabilities. While the performance of VLA models is promising, their computational cost can be substantial. This raises challenge for applying them on robotics tasks, which requires real-time decision-making to respond quickly to environmental changes. Since robotic control involves sequential decision-making, the visual input often exhibits minimal variation between successive steps. A natural idea is to reuse the computational results of unchanged visual tokens from the last step. Motivated by this idea, we propose VLA-Cache, an efficient vision-language-action model. VLA-Cache incorporates a token-selection mechanism that compares the visual input at each step with the input from the previous step, adaptively identifying visual tokens with minimal changes. The computational results for these unchanged tokens are then reused in subsequent steps via KV-cache, thereby significantly improving the efficiency of the VLA-Cache model. Experimental results on both simulation (e.g., LIBERO benchmark and SIMPLER) and real-world robot valid VLA-Cache can achieve practical acceleration with minimal sacrifice in success rate.

Color Correction Meets Cross-Spectral Refinement: A Distribution-Aware Diffusion for Underwater Image Restoration

Underwater imaging often suffers from significant visual degradation, which limits its suitability for subsequent applications. While recent underwater image enhancement (UIE) methods rely on the current advances in deep neural network architecture designs, there is still considerable room for improvement in terms of cross-scene robustness and computational efficiency. Diffusion models have shown great success in image generation, prompting us to consider their application to UIE tasks. However, directly applying them to UIE tasks will pose two challenges, \textit{i.e.}, high computational budget and color unbalanced perturbations. To tackle these issues, we propose DiffColor, a distribution-aware diffusion and cross-spectral refinement model for efficient UIE. Instead of diffusing in the raw pixel space, we transfer the image into the wavelet domain to obtain such low-frequency and high-frequency spectra, it inherently reduces the image spatial dimensions by half after each transformation. Unlike single-noise image restoration tasks, underwater imaging exhibits unbalanced channel distributions due to the selective absorption of light by water. To address this, we design the Global Color Correction (GCC) module to handle the diverse color shifts, thereby avoiding potential global degradation disturbances during the denoising process. For the sacrificed image details caused by underwater scattering, we further present the Cross-Spectral Detail Refinement (CSDR) to enhance the high-frequency details, which are integrated with the low-frequency signal as input conditions for guiding the diffusion. This way not only ensures the high-fidelity of sampled content but also compensates for the sacrificed details. Comprehensive experiments demonstrate the superior performance of DiffColor over state-of-the-art methods in both quantitative and qualitative evaluations.

FusionSAM: Latent Space driven Segment Anything Model for Multimodal Fusion and Segmentation

Multimodal image fusion and segmentation enhance scene understanding in autonomous driving by integrating data from various sensors. However, current models struggle to efficiently segment densely packed elements in such scenes, due to the absence of comprehensive fusion features that can guide mid-process fine-tuning and focus attention on relevant areas. The Segment Anything Model (SAM) has emerged as a transformative segmentation method. It provides more effective prompts through its flexible prompt encoder, compared to transformers lacking fine-tuned control. Nevertheless, SAM has not been extensively studied in the domain of multimodal fusion for natural images. In this paper, we introduce SAM into multimodal image segmentation for the first time, proposing a novel framework that combines Latent Space Token Generation (LSTG) and Fusion Mask Prompting (FMP) modules to enhance SAM's multimodal fusion and segmentation capabilities. Specifically, we first obtain latent space features of the two modalities through vector quantization and embed them into a cross-attention-based inter-domain fusion module to establish long-range dependencies between modalities. Then, we use these comprehensive fusion features as prompts to guide precise pixel-level segmentation. Extensive experiments on several public datasets demonstrate that the proposed method significantly outperforms SAM and SAM2 in multimodal autonomous driving scenarios, achieving at least 3.9$\%$ higher segmentation mIoU than the state-of-the-art approaches.

Collage Prompting: Budget-Friendly Visual Recognition with GPT-4V

Recent advancements in generative AI have suggested that by taking visual prompt, GPT-4V can demonstrate significant proficiency in image recognition task. Despite its impressive capabilities, the financial cost associated with GPT-4V's inference presents a substantial barrier for its wide use. To address this challenge, our work introduces Collage Prompting, a budget-friendly prompting approach that concatenates multiple images into a single visual input. With collage prompt, GPT-4V is able to perform image recognition on several images simultaneously. Based on the observation that the accuracy of GPT-4V's image recognition varies significantly with the order of images within the collage prompt, our method further learns to optimize the arrangement of images for maximum recognition accuracy. A graph predictor is trained to indicate the accuracy of each collage prompt, then we propose an optimization method to navigate the search space of possible image arrangements. Experiment results across various datasets demonstrate the cost-efficiency score of collage prompt is much larger than standard prompt. Additionally, collage prompt with learned arrangement achieves clearly better accuracy than collage prompt with random arrangement in GPT-4V's visual recognition.

Visual Imitation Learning with Calibrated Contrastive Representation

Adversarial Imitation Learning (AIL) allows the agent to reproduce expert behavior with low-dimensional states and actions. However, challenges arise in handling visual states due to their less distinguishable representation compared to low-dimensional proprioceptive features. While existing methods resort to adopt complex network architectures or separate the process of learning representation and decision-making, they overlook valuable intra-agent information within demonstrations. To address this problem, this paper proposes a simple and effective solution by incorporating calibrated contrastive representative learning into visual AIL framework. Specifically, we present an image encoder in visual AIL, utilizing a combination of unsupervised and supervised contrastive learning to extract valuable features from visual states. Based on the fact that the improved agent often produces demonstrations of varying quality, we propose to calibrate the contrastive loss by treating each agent demonstrations as a mixed sample. The incorporation of contrastive learning can be jointly optimized with the AIL framework, without modifying the architecture or incurring significant computational costs. Experimental results on DMControl Suite demonstrate our proposed method is sample efficient and can outperform other compared methods from different aspects.