Conflict-Aware Additive Guidance for Flow Models under Compositional Rewards

Inference-time guided sampling steers state-of-the-art diffusion and flow models without fine-tuning by interpreting the generation process as a controllable trajectory. This provides a simple and flexible way to inject external constraints (e.g., cost functions or pre-trained verifiers) for controlled generation. However, existing methods often fail when composing multiple constraints simultaneously, which leads to deviations from the true data manifold. In this work, we identify root causes of this off-manifold drift and find that the approximation error scales severely with gradient misalignment. Building on these findings, we propose Conflict-Aware Additive Guidance ($g^\text{car}$), a lightweight and learnable method, which actively rectifies off-manifold drift by dynamically detecting and resolving gradient conflicts. We validate $g^\text{car}$ across diverse domains, ranging from synthetic datasets and image editing to generative decision-making for planning and control. Our results demonstrate that $g^\text{car}$ effectively rectifies off-manifold drift, surpassing baselines in generation fidelity while using light compute. Code is available at https://github.com/yuxuehui/CAR-guidance.

Guided Streaming Stochastic Interpolant Policy

Inference-time guidance is essential for steering generative robot policies toward dynamic objectives without retraining, yet existing methods are largely confined to chunk-based architectures that exhibit high latency and lack the reactivity needed for test-time preference alignment or obstacle avoidance. In this work, we formally derive the optimal guidance term for Stochastic Interpolants (SI) by analyzing the value function's time evolution via the Backward Kolmogorov Equation, establishing a modified drift that theoretically guarantees sampling from a target distribution. We apply this framework to real-time control through the Streaming Stochastic Interpolant Policy (SSIP), which generalizes the deterministic Streaming Flow Policy (SFP). Unifying this guidance law with the streaming architecture enables fast and reactive control. To support diverse deployment needs, we propose two complementary mechanisms: training-free Stochastic Trajectory Ensemble Guidance (STEG) that computes gradients on-the-fly for zero-shot adaptation, and training-based Conditional Critic Guidance (CCG) for amortized inference. Empirical evaluations demonstrate that our guided streaming approach significantly outperforms conventional chunk-based policies in reactivity and provides superior, physically valid guidance for dynamic, unstructured environments.

Change-Robust Online Spatial-Semantic Topological Mapping

Autonomous robots require change-robust spatial-semantic reasoning: using spatial and semantic knowledge to decide where to go, how to get there, and where the robot is despite environmental change. Existing approaches typically attach semantics to SLAM-built metric maps, but these pipelines are brittle under appearance shifts and scene dynamics, where data association and relocalization degrade. We propose a Change-Robust Online Spatial-Semantic (CROSS) representation that replaces a globally consistent metric substrate with an online, pose-aware topological graph of RGB-D keyframes. The system explicitly reasons over perceptual ambiguity using sequential hypothesis testing in continuous SE(3). Our estimator maintains a bounded Gaussian-mixture belief over poses, enabling principled handling of loop closures and kidnapped-robot events. Experiments under severe appearance change, including real-robot object-goal navigation with lighting shifts and furniture rearrangement, demonstrate improved robustness over SLAM-based and topological baselines while remaining safe under perceptual aliasing.

Grounded World Model for Semantically Generalizable Planning

In Model Predictive Control (MPC), world models predict the future outcomes of various action proposals, which are then scored to guide the selection of the optimal action. For visuomotor MPC, the score function is a distance metric between a predicted image and a goal image, measured in the latent space of a pretrained vision encoder like DINO and JEPA. However, it is challenging to obtain the goal image in advance of the task execution, particularly in new environments. Additionally, conveying the goal through an image offers limited interactivity compared with natural language. In this work, we propose to learn a Grounded World Model (GWM) in a vision-language-aligned latent space. As a result, each proposed action is scored based on how close its future outcome is to the task instruction, reflected by the similarity of embeddings. This approach transforms the visuomotor MPC to a VLA that surpasses VLM-based VLAs in semantic generalization. On the proposed WISER benchmark, GWM-MPC achieves a 87% success rate on the test set comprising 288 tasks that feature unseen visual signals and referring expressions, yet remain solvable with motions demonstrated during training. In contrast, traditional VLAs achieve an average success rate of 22%, even though they overfit the training set with a 90% success rate.

SkillVLA: Tackling Combinatorial Diversity in Dual-Arm Manipulation via Skill Reuse

Recent progress in vision-language-action (VLA) models has demonstrated strong potential for dual-arm manipulation, enabling complex behaviors and generalization to unseen environments. However, mainstream bimanual VLA formulations largely overlook the critical challenge of combinatorial diversity. Different pairings of single-arm behaviors can induce qualitatively distinct task behaviors, yet existing models do not explicitly account for this structure. We argue that effective bimanual VLAs should support skill reuse - the ability to recombine previously learned single-arm skills across novel left-right pairings - thereby avoiding the need to separately learn every possible combination. Current VLA designs entangle skills across arms, preventing such recomposition and limiting scalability. To address this limitation, we propose SkillVLA, a framework explicitly designed to enable skill reuse in dual-arm manipulation. Extensive experiments demonstrate that SkillVLA substantially improves skill composition, increasing overall success rate from 0% to 51%, and achieves strong performance on cooperative and long-horizon tasks.

Action Hallucination in Generative Visual-Language-Action Models

Robot Foundation Models such as Vision-Language-Action models are rapidly reshaping how robot policies are trained and deployed, replacing hand-designed planners with end-to-end generative action models. While these systems demonstrate impressive generalization, it remains unclear whether they fundamentally resolve the long-standing challenges of robotics. We address this question by analyzing action hallucinations that violate physical constraints and their extension to plan-level failures. Focusing on latent-variable generative policies, we show that hallucinations often arise from structural mismatches between feasible robot behavior and common model architectures. We study three such barriers -- topological, precision, and horizon -- and show how they impose unavoidable tradeoffs. Our analysis provides mechanistic explanations for reported empirical failures of generative robot policies and suggests principled directions for improving reliability and trustworthiness, without abandoning their expressive power.

Not All Layers Need Tuning: Selective Layer Restoration Recovers Diversity

Post-training improves instruction-following and helpfulness of large language models (LLMs) but often reduces generation diversity, which leads to repetitive outputs in open-ended settings, a phenomenon known as mode collapse. Motivated by evidence that LLM layers play distinct functional roles, we hypothesize that mode collapse can be localized to specific layers and that restoring a carefully chosen range of layers to their pre-trained weights can recover diversity while maintaining high output quality. To validate this hypothesis and decide which layers to restore, we design a proxy task -- Constrained Random Character(CRC) -- with an explicit validity set and a natural diversity objective. Results on CRC reveal a clear diversity-validity trade-off across restoration ranges and identify configurations that increase diversity with minimal quality loss. Based on these findings, we propose Selective Layer Restoration (SLR), a training-free method that restores selected layers in a post-trained model to their pre-trained weights, yielding a hybrid model with the same architecture and parameter count, incurring no additional inference cost. Across three different tasks (creative writing, open-ended question answering, and multi-step reasoning) and three different model families (Llama, Qwen, and Gemma), we find SLR can consistently and substantially improve output diversity while maintaining high output quality.

Abstracting Robot Manipulation Skills via Mixture-of-Experts Diffusion Policies

Diffusion-based policies have recently shown strong results in robot manipulation, but their extension to multi-task scenarios is hindered by the high cost of scaling model size and demonstrations. We introduce Skill Mixture-of-Experts Policy (SMP), a diffusion-based mixture-of-experts policy that learns a compact orthogonal skill basis and uses sticky routing to compose actions from a small, task-relevant subset of experts at each step. A variational training objective supports this design, and adaptive expert activation at inference yields fast sampling without oversized backbones. We validate SMP in simulation and on a real dual-arm platform with multi-task learning and transfer learning tasks, where SMP achieves higher success rates and markedly lower inference cost than large diffusion baselines. These results indicate a practical path toward scalable, transferable multi-task manipulation: learn reusable skills once, activate only what is needed, and adapt quickly when tasks change.

OpenRoboCare: A Multimodal Multi-Task Expert Demonstration Dataset for Robot Caregiving

We present OpenRoboCare, a multimodal dataset for robot caregiving, capturing expert occupational therapist demonstrations of Activities of Daily Living (ADLs). Caregiving tasks involve complex physical human-robot interactions, requiring precise perception under occlusions, safe physical contact, and long-horizon planning. While recent advances in robot learning from demonstrations have shown promise, there is a lack of a large-scale, diverse, and expert-driven dataset that captures real-world caregiving routines. To address this gap, we collect data from 21 occupational therapists performing 15 ADL tasks on two manikins. The dataset spans five modalities: RGB-D video, pose tracking, eye-gaze tracking, task and action annotations, and tactile sensing, providing rich multimodal insights into caregiver movement, attention, force application, and task execution strategies. We further analyze expert caregiving principles and strategies, offering insights to improve robot efficiency and task feasibility. Additionally, our evaluations demonstrate that OpenRoboCare presents challenges for state-of-the-art robot perception and human activity recognition methods, both critical for developing safe and adaptive assistive robots, highlighting the value of our contribution. See our website for additional visualizations: https://emprise.cs.cornell.edu/robo-care/.

VLA-Touch: Enhancing Vision-Language-Action Models with Dual-Level Tactile Feedback

Tactile feedback is generally recognized to be crucial for effective interaction with the physical world. However, state-of-the-art Vision-Language-Action (VLA) models lack the ability to interpret and use tactile signals, limiting their effectiveness in contact-rich tasks. Incorporating tactile feedback into these systems is challenging due to the absence of large multi-modal datasets. We present VLA-Touch, an approach that enhances generalist robot policies with tactile sensing \emph{without fine-tuning} the base VLA. Our method introduces two key innovations: (1) a pipeline that leverages a pretrained tactile-language model that provides semantic tactile feedback for high-level task planning, and (2) a diffusion-based controller that refines VLA-generated actions with tactile signals for contact-rich manipulation. Through real-world experiments, we demonstrate that our dual-level integration of tactile feedback improves task planning efficiency while enhancing execution precision. Code is open-sourced at \href{https://github.com/jxbi1010/VLA-Touch}{this URL}.