Xueqian Wang

Yifan Shi, Yingqi Liu, Kang Wei et al.

To defend the inference attacks and mitigate the sensitive information leakages in Federated Learning (FL), client-level Differentially Private FL (DPFL) is the de-facto standard for privacy protection by clipping local updates and adding random noise. However, existing DPFL methods tend to make a sharper loss landscape and have poorer weight perturbation robustness, resulting in severe performance degradation. To alleviate these issues, we propose a novel DPFL algorithm named DP-FedSAM, which leverages gradient perturbation to mitigate the negative impact of DP. Specifically, DP-FedSAM integrates Sharpness Aware Minimization (SAM) optimizer to generate local flatness models with better stability and weight perturbation robustness, which results in the small norm of local updates and robustness to DP noise, thereby improving the performance. From the theoretical perspective, we analyze in detail how DP-FedSAM mitigates the performance degradation induced by DP. Meanwhile, we give rigorous privacy guarantees with Rényi DP and present the sensitivity analysis of local updates. At last, we empirically confirm that our algorithm achieves state-of-the-art (SOTA) performance compared with existing SOTA baselines in DPFL. Code is available at https://github.com/YMJS-Irfan/DP-FedSAM

Linrui Zhang, Qin Zhang, Li Shen et al.

Safe reinforcement learning (RL) has achieved significant success on risk-sensitive tasks and shown promise in autonomous driving (AD) as well. Considering the distinctiveness of this community, efficient and reproducible baselines are still lacking for safe AD. In this paper, we release SafeRL-Kit to benchmark safe RL methods for AD-oriented tasks. Concretely, SafeRL-Kit contains several latest algorithms specific to zero-constraint-violation tasks, including Safety Layer, Recovery RL, off-policy Lagrangian method, and Feasible Actor-Critic. In addition to existing approaches, we propose a novel first-order method named Exact Penalty Optimization (EPO) and sufficiently demonstrate its capability in safe AD. All algorithms in SafeRL-Kit are implemented (i) under the off-policy setting, which improves sample efficiency and can better leverage past logs; (ii) with a unified learning framework, providing off-the-shelf interfaces for researchers to incorporate their domain-specific knowledge into fundamental safe RL methods. Conclusively, we conduct a comparative evaluation of the above algorithms in SafeRL-Kit and shed light on their efficacy for safe autonomous driving. The source code is available at \href{ https://github.com/zlr20/saferl_kit}{this https URL}.

Haoyu Wang, Guozheng Ma, Cong Yu et al.

The swift advancement in the scales and capabilities of Large Language Models (LLMs) positions them as promising tools for a variety of downstream tasks. In addition to the pursuit of better performance and the avoidance of violent feedback on a certain prompt, to ensure the responsibility of the LLM, much attention is drawn to the robustness of LLMs. However, existing evaluation methods mostly rely on traditional question answering datasets with predefined supervised labels, which do not align with the superior generation capabilities of contemporary LLMs. To address this issue, we propose a novel rational evaluation approach that leverages pre-trained reward models as diagnostic tools to evaluate the longer conversation generated from more challenging open questions by LLMs, which we refer to as the Reward Model for Reasonable Robustness Evaluation (TREvaL). Longer conversations manifest the comprehensive grasp of language models in terms of their proficiency in understanding questions, a capability not entirely encompassed by individual words or letters, which may exhibit oversimplification and inherent biases. Our extensive empirical experiments demonstrate that TREvaL provides an innovative method for evaluating the robustness of an LLM. Furthermore, our results demonstrate that LLMs frequently exhibit vulnerability to word-level perturbations that are commonplace in daily language usage. Notably, we are surprised to discover that robustness tends to decrease as fine-tuning (SFT and RLHF) is conducted. The code of TREval is available in https://github.com/Harry-mic/TREvaL.

Zihao Lin, Yan Sun, Yifan Shi et al.

With the blowout development of pre-trained models (PTMs), the efficient tuning of these models for diverse downstream applications has emerged as a pivotal research concern. Although recent investigations into prompt tuning have provided promising avenues, three salient challenges persist: (1) memory constraint: the continuous growth in the size of open-source PTMs renders fine-tuning, even a fraction of their parameters, challenging for many practitioners. (2) model privacy: existing PTMs often function as public API services, with their parameters inaccessible for effective or tailored fine-tuning. (3) data privacy: the fine-tuning of PTMs necessitates high-quality datasets, which are typically localized and not shared to public. To optimally harness each local dataset while navigating memory constraints and preserving privacy, we propose Federated Black-Box Prompt Tuning (Fed-BBPT). This innovative approach eschews reliance on parameter architectures and private dataset access, instead capitalizing on a central server that aids local users in collaboratively training a prompt generator through regular aggregation. Local users leverage API-driven learning via a zero-order optimizer, obviating the need for PTM deployment. Relative to extensive fine-tuning, Fed-BBPT proficiently sidesteps memory challenges tied to PTM storage and fine-tuning on local machines, tapping into comprehensive, high-quality, yet private training datasets. A thorough evaluation across 40 datasets spanning CV and NLP tasks underscores the robustness of our proposed model.

Yihao Wu, He Zhang, Junbo Tan et al.

Post-training Vision-Language-Action (VLA) models into policies that can be reliably deployed on real robots remains a major bottleneck. SFT and DAgger exploit failure signals only indirectly, and reward-based RL is bottlenecked by the difficulty of real-world reward design and of training reliable critics. We present FlowPRO, a reward-free offline reinforced fine-tuning framework for flow-matching VLAs. Algorithmically, we propose RPRO (Robotic Flow-matching Proximalized Preference Optimization), a preference-optimization objective tailored to the flow-matching action head of VLA models. RPRO pairs a contrastive optimizer with an explicit proximal regularizer that anchors the absolute magnitude of the implicit reward, thereby eliminating the reward-hacking failure mode of plain Flow-DPO. On the data side, a teleoperated intervention-and-rollback paradigm produces naturally paired positive and negative trajectories $(τ^w, τ^l)$ on a real robot from a single operator action; a Smooth Interpolation procedure, combined with batch mixing, then converts these sparse corrections into dense per-state supervision while preserving the base policy's capabilities. On four long-horizon bimanual tasks, FlowPRO attains the highest success rate, outperforming four representative baselines, and ablations confirm the contribution of each loss component.

Yilun Kong, Hangyu Mao, Qi Zhao et al.

Prompt engineering has demonstrated remarkable success in enhancing the performance of large language models (LLMs) across diverse tasks. However, most existing prompt optimization methods only focus on the task-level performance, overlooking the importance of query-preferred prompts, which leads to suboptimal performances. Additionally, these methods rely heavily on frequent interactions with LLMs to obtain feedback for guiding the optimization process, incurring substantial redundant interaction costs. In this paper, we introduce Query-dependent Prompt Optimization (QPO), which leverages multi-loop offline reinforcement learning to iteratively fine-tune a small pretrained language model to generate optimal prompts tailored to the input queries, thus significantly improving the prompting effect on the large target LLM. We derive insights from offline prompting demonstration data, which already exists in large quantities as a by-product of benchmarking diverse prompts on open-sourced tasks, thereby circumventing the expenses of online interactions. Furthermore, we continuously augment the offline dataset with the generated prompts in each loop, as the prompts from the fine-tuned model are supposed to outperform the source prompts in the original dataset. These iterative loops bootstrap the model towards generating optimal prompts. Experiments on various LLM scales and diverse NLP and math tasks demonstrate the efficacy and cost-efficiency of our method in both zero-shot and few-shot scenarios.

Longxiang He, Li Shen, Linrui Zhang et al.

Constrained policy search (CPS) is a fundamental problem in offline reinforcement learning, which is generally solved by advantage weighted regression (AWR). However, previous methods may still encounter out-of-distribution actions due to the limited expressivity of Gaussian-based policies. On the other hand, directly applying the state-of-the-art models with distribution expression capabilities (i.e., diffusion models) in the AWR framework is intractable since AWR requires exact policy probability densities, which is intractable in diffusion models. In this paper, we propose a novel approach, $\textbf{Diffusion-based Constrained Policy Search}$ (dubbed DiffCPS), which tackles the diffusion-based constrained policy search with the primal-dual method. The theoretical analysis reveals that strong duality holds for diffusion-based CPS problems, and upon introducing parameter approximation, an approximated solution can be obtained after $\mathcal{O}(1/ε)$ number of dual iterations, where $ε$ denotes the representation ability of the parametrized policy. Extensive experimental results based on the D4RL benchmark demonstrate the efficacy of our approach. We empirically show that DiffCPS achieves better or at least competitive performance compared to traditional AWR-based baselines as well as recent diffusion-based offline RL methods. The code is now available at https://github.com/felix-thu/DiffCPS.

Mingqi Shao, Chongkun Xia, Dongxu Duan et al.

In this work, we propose a novel method for the detailed reconstruction of transparent objects by exploiting polarimetric cues. Most of the existing methods usually lack sufficient constraints and suffer from the over-smooth problem. Hence, we introduce polarization information as a complementary cue. We implicitly represent the object's geometry as a neural network, while the polarization render is capable of rendering the object's polarization images from the given shape and illumination configuration. Direct comparison of the rendered polarization images to the real-world captured images will have additional errors due to the transmission in the transparent object. To address this issue, the concept of reflection percentage which represents the proportion of the reflection component is introduced. The reflection percentage is calculated by a ray tracer and then used for weighting the polarization loss. We build a polarization dataset for multi-view transparent shapes reconstruction to verify our method. The experimental results show that our method is capable of recovering detailed shapes and improving the reconstruction quality of transparent objects. Our dataset and code will be publicly available at https://github.com/shaomq2187/TransPIR.

Linrui Zhang, Qin Zhang, Li Shen et al.

Safety comes first in many real-world applications involving autonomous agents. Despite a large number of reinforcement learning (RL) methods focusing on safety-critical tasks, there is still a lack of high-quality evaluation of those algorithms that adheres to safety constraints at each decision step under complex and unknown dynamics. In this paper, we revisit prior work in this scope from the perspective of state-wise safe RL and categorize them as projection-based, recovery-based, and optimization-based approaches, respectively. Furthermore, we propose Unrolling Safety Layer (USL), a joint method that combines safety optimization and safety projection. This novel technique explicitly enforces hard constraints via the deep unrolling architecture and enjoys structural advantages in navigating the trade-off between reward improvement and constraint satisfaction. To facilitate further research in this area, we reproduce related algorithms in a unified pipeline and incorporate them into SafeRL-Kit, a toolkit that provides off-the-shelf interfaces and evaluation utilities for safety-critical tasks. We then perform a comparative study of the involved algorithms on six benchmarks ranging from robotic control to autonomous driving. The empirical results provide an insight into their applicability and robustness in learning zero-cost-return policies without task-dependent handcrafting. The project page is available at https://sites.google.com/view/saferlkit.

Linrui Zhang, Li Shen, Long Yang et al.

Safe reinforcement learning aims to learn the optimal policy while satisfying safety constraints, which is essential in real-world applications. However, current algorithms still struggle for efficient policy updates with hard constraint satisfaction. In this paper, we propose Penalized Proximal Policy Optimization (P3O), which solves the cumbersome constrained policy iteration via a single minimization of an equivalent unconstrained problem. Specifically, P3O utilizes a simple-yet-effective penalty function to eliminate cost constraints and removes the trust-region constraint by the clipped surrogate objective. We theoretically prove the exactness of the proposed method with a finite penalty factor and provide a worst-case analysis for approximate error when evaluated on sample trajectories. Moreover, we extend P3O to more challenging multi-constraint and multi-agent scenarios which are less studied in previous work. Extensive experiments show that P3O outperforms state-of-the-art algorithms with respect to both reward improvement and constraint satisfaction on a set of constrained locomotive tasks.

Yifan Shi, Li Shen, Kang Wei et al.

To mitigate the privacy leakages and communication burdens of Federated Learning (FL), decentralized FL (DFL) discards the central server and each client only communicates with its neighbors in a decentralized communication network. However, existing DFL suffers from high inconsistency among local clients, which results in severe distribution shift and inferior performance compared with centralized FL (CFL), especially on heterogeneous data or sparse communication topology. To alleviate this issue, we propose two DFL algorithms named DFedSAM and DFedSAM-MGS to improve the performance of DFL. Specifically, DFedSAM leverages gradient perturbation to generate local flat models via Sharpness Aware Minimization (SAM), which searches for models with uniformly low loss values. DFedSAM-MGS further boosts DFedSAM by adopting Multiple Gossip Steps (MGS) for better model consistency, which accelerates the aggregation of local flat models and better balances communication complexity and generalization. Theoretically, we present improved convergence rates $\small \mathcal{O}\big(\frac{1}{\sqrt{KT}}+\frac{1}{T}+\frac{1}{K^{1/2}T^{3/2}(1-λ)^2}\big)$ and $\small \mathcal{O}\big(\frac{1}{\sqrt{KT}}+\frac{1}{T}+\frac{λ^Q+1}{K^{1/2}T^{3/2}(1-λ^Q)^2}\big)$ in non-convex setting for DFedSAM and DFedSAM-MGS, respectively, where $1-λ$ is the spectral gap of gossip matrix and $Q$ is the number of MGS. Empirically, our methods can achieve competitive performance compared with CFL methods and outperform existing DFL methods.

Qin Zhang, Linrui Zhang, Haoran Xu et al.

Offline safe RL is of great practical relevance for deploying agents in real-world applications. However, acquiring constraint-satisfying policies from the fixed dataset is non-trivial for conventional approaches. Even worse, the learned constraints are stationary and may become invalid when the online safety requirement changes. In this paper, we present a novel offline safe RL approach referred to as SaFormer, which tackles the above issues via conditional sequence modeling. In contrast to existing sequence models, we propose cost-related tokens to restrict the action space and a posterior safety verification to enforce the constraint explicitly. Specifically, SaFormer performs a two-stage auto-regression conditioned by the maximum remaining cost to generate feasible candidates. It then filters out unsafe attempts and executes the optimal action with the highest expected return. Extensive experiments demonstrate the efficacy of SaFormer featuring (1) competitive returns with tightened constraint satisfaction; (2) adaptability to the in-range cost values of the offline data without retraining; (3) generalizability for constraints beyond the current dataset.

Weiming Li, Xueqian Wang, Gang Li et al.

Change detection (CD) in heterogeneous remote sensing images has been widely used for disaster monitoring and land-use management. In the past decade, the heterogeneous CD problem has significantly benefited from the development of deep neural networks (DNNs). However, the purely data-driven DNNs perform like a black box where the lack of interpretability limits the trustworthiness and controllability of DNNs in most practical CD applications. As a powerful knowledge-driven tool, copula theory performs well in modeling relationships among random variables. To enhance the interpretability of existing neural networks for CD, we propose a knowledge-data-driven heterogeneous CD method based on a copula-guided neural network, named NN-Copula-CD. In our NN-Copula-CD, the mathematical characteristics of copula are employed as the loss functions to supervise a neural network to learn the dependence between bi-temporal heterogeneous superpixel pairs, and then the changed regions are identified via binary classification based on the degrees of dependence of all the superpixel pairs in the bi-temporal images. We conduct in-depth experiments on three datasets with heterogeneous images, where both quantitative and visual results demonstrate the effectiveness of our proposed NN-Copula-CD method.

Guozheng Ma, Lu Li, Sen Zhang et al.

Plasticity, the ability of a neural network to evolve with new data, is crucial for high-performance and sample-efficient visual reinforcement learning (VRL). Although methods like resetting and regularization can potentially mitigate plasticity loss, the influences of various components within the VRL framework on the agent's plasticity are still poorly understood. In this work, we conduct a systematic empirical exploration focusing on three primary underexplored facets and derive the following insightful conclusions: (1) data augmentation is essential in maintaining plasticity; (2) the critic's plasticity loss serves as the principal bottleneck impeding efficient training; and (3) without timely intervention to recover critic's plasticity in the early stages, its loss becomes catastrophic. These insights suggest a novel strategy to address the high replay ratio (RR) dilemma, where exacerbated plasticity loss hinders the potential improvements of sample efficiency brought by increased reuse frequency. Rather than setting a static RR for the entire training process, we propose Adaptive RR, which dynamically adjusts the RR based on the critic's plasticity level. Extensive evaluations indicate that Adaptive RR not only avoids catastrophic plasticity loss in the early stages but also benefits from more frequent reuse in later phases, resulting in superior sample efficiency.

Zhecheng Yuan, Zhengrong Xue, Bo Yuan et al.

Learning generalizable policies that can adapt to unseen environments remains challenging in visual Reinforcement Learning (RL). Existing approaches try to acquire a robust representation via diversifying the appearances of in-domain observations for better generalization. Limited by the specific observations of the environment, these methods ignore the possibility of exploring diverse real-world image datasets. In this paper, we investigate how a visual RL agent would benefit from the off-the-shelf visual representations. Surprisingly, we find that the early layers in an ImageNet pre-trained ResNet model could provide rather generalizable representations for visual RL. Hence, we propose Pre-trained Image Encoder for Generalizable visual reinforcement learning (PIE-G), a simple yet effective framework that can generalize to the unseen visual scenarios in a zero-shot manner. Extensive experiments are conducted on DMControl Generalization Benchmark, DMControl Manipulation Tasks, Drawer World, and CARLA to verify the effectiveness of PIE-G. Empirical evidence suggests PIE-G improves sample efficiency and significantly outperforms previous state-of-the-art methods in terms of generalization performance. In particular, PIE-G boasts a 55% generalization performance gain on average in the challenging video background setting. Project Page: https://sites.google.com/view/pie-g/home.

Guozheng Ma, Zhen Wang, Zhecheng Yuan et al.

Visual reinforcement learning (RL), which makes decisions directly from high-dimensional visual inputs, has demonstrated significant potential in various domains. However, deploying visual RL techniques in the real world remains challenging due to their low sample efficiency and large generalization gaps. To tackle these obstacles, data augmentation (DA) has become a widely used technique in visual RL for acquiring sample-efficient and generalizable policies by diversifying the training data. This survey aims to provide a timely and essential review of DA techniques in visual RL in recognition of the thriving development in this field. In particular, we propose a unified framework for analyzing visual RL and understanding the role of DA in it. We then present a principled taxonomy of the existing augmentation techniques used in visual RL and conduct an in-depth discussion on how to better leverage augmented data in different scenarios. Moreover, we report a systematic empirical evaluation of DA-based techniques in visual RL and conclude by highlighting the directions for future research. As the first comprehensive survey of DA in visual RL, this work is expected to offer valuable guidance to this emerging field.

Kai Mo, Chongkun Xia, Xueqian Wang et al.

Sequential multi-step cloth manipulation is a challenging problem in robotic manipulation, requiring a robot to perceive the cloth state and plan a sequence of chained actions leading to the desired state. Most previous works address this problem in a goal-conditioned way, and goal observation must be given for each specific task and cloth configuration, which is not practical and efficient. Thus, we present a novel multi-step cloth manipulation planning framework named Foldformer. Foldformer can complete similar tasks with only a general demonstration and utilize a space-time attention mechanism to capture the instruction information behind this demonstration. We experimentally evaluate Foldsformer on four representative sequential multi-step manipulation tasks and show that Foldsformer significantly outperforms state-of-the-art approaches in simulation. Foldformer can complete multi-step cloth manipulation tasks even when configurations of the cloth (e.g., size and pose) vary from configurations in the general demonstrations. Furthermore, our approach can be transferred from simulation to the real world without additional training or domain randomization. Despite training on rectangular clothes, we also show that our approach can generalize to unseen cloth shapes (T-shirts and shorts). Videos and source code are available at: https://sites.google.com/view/foldsformer.

Tiantian Zhang, Zichuan Lin, Yuxing Wang et al.

A key challenge of continual reinforcement learning (CRL) in dynamic environments is to promptly adapt the RL agent's behavior as the environment changes over its lifetime, while minimizing the catastrophic forgetting of the learned information. To address this challenge, in this article, we propose DaCoRL, i.e., dynamics-adaptive continual RL. DaCoRL learns a context-conditioned policy using progressive contextualization, which incrementally clusters a stream of stationary tasks in the dynamic environment into a series of contexts and opts for an expandable multihead neural network to approximate the policy. Specifically, we define a set of tasks with similar dynamics as an environmental context and formalize context inference as a procedure of online Bayesian infinite Gaussian mixture clustering on environment features, resorting to online Bayesian inference to infer the posterior distribution over contexts. Under the assumption of a Chinese restaurant process prior, this technique can accurately classify the current task as a previously seen context or instantiate a new context as needed without relying on any external indicator to signal environmental changes in advance. Furthermore, we employ an expandable multihead neural network whose output layer is synchronously expanded with the newly instantiated context, and a knowledge distillation regularization term for retaining the performance on learned tasks. As a general framework that can be coupled with various deep RL algorithms, DaCoRL features consistent superiority over existing methods in terms of the stability, overall performance and generalization ability, as verified by extensive experiments on several robot navigation and MuJoCo locomotion tasks.

Shoujie Li, Haixin Yu, Wenbo Ding et al.

The accurate detection and grasping of transparent objects are challenging but of significance to robots. Here, a visual-tactile fusion framework for transparent object grasping under complex backgrounds and variant light conditions is proposed, including the grasping position detection, tactile calibration, and visual-tactile fusion based classification. First, a multi-scene synthetic grasping dataset generation method with a Gaussian distribution based data annotation is proposed. Besides, a novel grasping network named TGCNN is proposed for grasping position detection, showing good results in both synthetic and real scenes. In tactile calibration, inspired by human grasping, a fully convolutional network based tactile feature extraction method and a central location based adaptive grasping strategy are designed, improving the success rate by 36.7% compared to direct grasping. Furthermore, a visual-tactile fusion method is proposed for transparent objects classification, which improves the classification accuracy by 34%. The proposed framework synergizes the advantages of vision and touch, and greatly improves the grasping efficiency of transparent objects.

Qinglong Meng, Chongkun Xia, Xueqian Wang

Inverse Kinematics (IK) is to find the robot's configurations that satisfy the target pose of the end effector. In motion planning, diverse configurations were required in case a feasible trajectory was not found. Meanwhile, collision checking (CC), e.g. Oriented bounding box (OBB), Discrete Oriented Polytope (DOP), and Quickhull \cite{quickhull}, needs to be done for each configuration provided by the IK solver to ensure every goal configuration for motion planning is available. This means the classical IK solver and CC algorithm should be executed repeatedly for every configuration. Thus, the preparation time is long when the required number of goal configurations is large, e.g. motion planning in cluster environments. Moreover, structured maps, which might be difficult to obtain, were required by classical collision-checking algorithms. To sidestep such two issues, we propose a flow-based vision method that can output diverse available configurations by fusing inverse kinematics and collision checking, named Vision Inverse Kinematics solver (ViIK). Moreover, ViIK uses RGB images as the perception of environments. ViIK can output 1000 configurations within 40 ms, and the accuracy is about 3 millimeters and 1.5 degrees. The higher accuracy can be obtained by being refined by the classical IK solver within a few iterations. The self-collision rates can be lower than 2%. The collision-with-env rates can be lower than 10% in most scenes. The code is available at: https://github.com/AdamQLMeng/ViIK.

Zixuan Liu, Liu Liu, Xueqian Wang et al.

Differentiable optimization has received a significant amount of attention due to its foundational role in the domain of machine learning based on neural networks. This paper proposes a differentiable layer, named Differentiable Frank-Wolfe Layer (DFWLayer), by rolling out the Frank-Wolfe method, a well-known optimization algorithm which can solve constrained optimization problems without projections and Hessian matrix computations, thus leading to an efficient way of dealing with large-scale convex optimization problems with norm constraints. Experimental results demonstrate that the DFWLayer not only attains competitive accuracy in solutions and gradients but also consistently adheres to constraints.

Yuxing Wang, Zhiyu Chen, Tiantian Zhang et al.

Brain-body co-evolution enables animals to develop complex behaviors in their environments. Inspired by this biological synergy, embodied co-design (ECD) has emerged as a transformative paradigm for creating intelligent agents-from virtual creatures to physical robots-by jointly optimizing their morphologies and controllers rather than treating control in isolation. This integrated approach facilitates richer environmental interactions and robust task performance. In this survey, we provide a systematic overview of recent advances in ECD. We first formalize the concept of ECD and position it within related fields. We then introduce a hierarchical taxonomy: a lower layer that breaks down agent design into three fundamental components-controlling brain, body morphology, and task environment-and an upper layer that integrates these components into four major ECD frameworks: bi-level, single-level, generative, and open-ended. This taxonomy allows us to synthesize insights from more than one hundred recent studies. We further review notable benchmarks, datasets, and applications in both simulated and real-world scenarios. Finally, we identify significant challenges and offer insights into promising future research directions. A project associated with this survey has been created at https://github.com/Yuxing-Wang-THU/SurveyBrainBody.

Jinghui Xu, Boyi Shangguan, Mengke Zhu et al.

Autonomous Earth Observation (EO) agents are transitioning from passive perception to complex, multi-step task execution. However, current architectures that integrate planning and execution within a single model often struggle with combinatorial complexity and reasoning errors in dynamic EO scenarios. To resolve these challenges, we propose the Lightweight Multimodal Meta-Planner (LMMP) framework. LMMP incorporates a dual-awareness mechanism that grounds strategic plans in both multimodal image features and high-level task semantics. Crucially, we introduce a Meta Task Library to inject remote sensing expert knowledge directly into the workflow, which standardizes domain logic and ensures plans are physically feasible. We further implement a two-stage training pipeline, initializing the Meta-Planner via expert-distilled Supervised Fine-Tuning and refining it through Direct Preference Optimization based on execution feedback. Extensive experiments on a dataset derived from EarthBench and ThinkGeo demonstrate that LMMP significantly improves tool-calling accuracy and task success rates. Moreover, the framework exhibits strong ``plug-and-play'' versatility, consistently enhancing the performance of diverse executor backbones across previously unseen EO missions.

Weiming Li, Lihui Xue, Xueqian Wang et al.

For the task of change detection (CD) in remote sensing images, deep convolution neural networks (CNNs)-based methods have recently aggregated transformer modules to improve the capability of global feature extraction. However, they suffer degraded CD performance on small changed areas due to the simple single-scale integration of deep CNNs and transformer modules. To address this issue, we propose a hybrid network based on multi-scale CNN-transformer structure, termed MCTNet, where the multi-scale global and local information are exploited to enhance the robustness of the CD performance on changed areas with different sizes. Especially, we design the ConvTrans block to adaptively aggregate global features from transformer modules and local features from CNN layers, which provides abundant global-local features with different scales. Experimental results demonstrate that our MCTNet achieves better detection performance than existing state-of-the-art CD methods.

Linrui Zhang, Zichen Yan, Li Shen et al.

Learning a risk-aware policy is essential but rather challenging in unstructured robotic tasks. Safe reinforcement learning methods open up new possibilities to tackle this problem. However, the conservative policy updates make it intractable to achieve sufficient exploration and desirable performance in complex, sample-expensive environments. In this paper, we propose a dual-agent safe reinforcement learning strategy consisting of a baseline and a safe agent. Such a decoupled framework enables high flexibility, data efficiency and risk-awareness for RL-based control. Concretely, the baseline agent is responsible for maximizing rewards under standard RL settings. Thus, it is compatible with off-the-shelf training techniques of unconstrained optimization, exploration and exploitation. On the other hand, the safe agent mimics the baseline agent for policy improvement and learns to fulfill safety constraints via off-policy RL tuning. In contrast to training from scratch, safe policy correction requires significantly fewer interactions to obtain a near-optimal policy. The dual policies can be optimized synchronously via a shared replay buffer, or leveraging the pre-trained model or the non-learning-based controller as a fixed baseline agent. Experimental results show that our approach can learn feasible skills without prior knowledge as well as deriving risk-averse counterparts from pre-trained unsafe policies. The proposed method outperforms the state-of-the-art safe RL algorithms on difficult robot locomotion and manipulation tasks with respect to both safety constraint satisfaction and sample efficiency.

Zhengrong Xue, Zhecheng Yuan, Jiashun Wang et al.

Can a robot manipulate intra-category unseen objects in arbitrary poses with the help of a mere demonstration of grasping pose on a single object instance? In this paper, we try to address this intriguing challenge by using USEEK, an unsupervised SE(3)-equivariant keypoints method that enjoys alignment across instances in a category, to perform generalizable manipulation. USEEK follows a teacher-student structure to decouple the unsupervised keypoint discovery and SE(3)-equivariant keypoint detection. With USEEK in hand, the robot can infer the category-level task-relevant object frames in an efficient and explainable manner, enabling manipulation of any intra-category objects from and to any poses. Through extensive experiments, we demonstrate that the keypoints produced by USEEK possess rich semantics, thus successfully transferring the functional knowledge from the demonstration object to the novel ones. Compared with other object representations for manipulation, USEEK is more adaptive in the face of large intra-category shape variance, more robust with limited demonstrations, and more efficient at inference time.

Tiantian Zhang, Kevin Zehua Shen, Zichuan Lin et al.

Replaying past experiences has proven to be a highly effective approach for averting catastrophic forgetting in supervised continual learning. However, some crucial factors are still largely ignored, making it vulnerable to serious failure, when used as a solution to forgetting in continual reinforcement learning, even in the context of perfect memory where all data of previous tasks are accessible in the current task. On the one hand, since most reinforcement learning algorithms are not invariant to the reward scale, the previously well-learned tasks (with high rewards) may appear to be more salient to the current learning process than the current task (with small initial rewards). This causes the agent to concentrate on those salient tasks at the expense of generality on the current task. On the other hand, offline learning on replayed tasks while learning a new task may induce a distributional shift between the dataset and the learned policy on old tasks, resulting in forgetting. In this paper, we introduce RECALL, a replay-enhanced method that greatly improves the plasticity of existing replay-based methods on new tasks while effectively avoiding the recurrence of catastrophic forgetting in continual reinforcement learning. RECALL leverages adaptive normalization on approximate targets and policy distillation on old tasks to enhance generality and stability, respectively. Extensive experiments on the Continual World benchmark show that RECALL performs significantly better than purely perfect memory replay, and achieves comparable or better overall performance against state-of-the-art continual learning methods.

Yuhong Deng, Chongkun Xia, Xueqian Wang et al.

Object rearranging is one of the most common deformable manipulation tasks, where the robot needs to rearrange a deformable object into a goal configuration. Previous studies focus on designing an expert system for each specific task by model-based or data-driven approaches and the application scenarios are therefore limited. Some research has been attempting to design a general framework to obtain more advanced manipulation capabilities for deformable rearranging tasks, with lots of progress achieved in simulation. However, transferring from simulation to reality is difficult due to the limitation of the end-to-end CNN architecture. To address these challenges, we design a local GNN (Graph Neural Network) based learning method, which utilizes two representation graphs to encode keypoints detected from images. Self-attention is applied for graph updating and cross-attention is applied for generating manipulation actions. Extensive experiments have been conducted to demonstrate that our framework is effective in multiple 1-D (rope, rope ring) and 2-D (cloth) rearranging tasks in simulation and can be easily transferred to a real robot by fine-tuning a keypoint detector.

Mingqi Shao, Chongkun Xia, Zhendong Yang et al.

This paper presents a learning-based method for transparent surface estimation from a single view polarization image. Existing shape from polarization(SfP) methods have the difficulty in estimating transparent shape since the inherent transmission interference heavily reduces the reliability of physics-based prior. To address this challenge, we propose the concept of physics-based prior, which is inspired by the characteristic that the transmission component in the polarization image has more noise than reflection. The confidence is used to determine the contribution of the interfered physics-based prior. Then, we build a network(TransSfP) with multi-branch architecture to avoid the destruction of relationships between different hierarchical inputs. To train and test our method, we construct a dataset for transparent shape from polarization with paired polarization images and ground-truth normal maps. Extensive experiments and comparisons demonstrate the superior accuracy of our method.

Lirong Che, Yuzhe yang, Peiwen lin et al.

Agent harness evolution improves frozen language-model agents by modifying the executable structures around them. We study this paradigm as a form of sample-efficient fast adaptation: instead of updating model weights, an agent can acquire task-specific competence by changing its external harness, while leaving the base model's general capabilities intact. Prior work shows that self-generated rollouts can support harness search, suggesting that agents may acquire new task competence through practice. Yet in long-horizon stochastic environments, self-practice becomes fragile: rewards are sparse, outcomes are high-variance, and failures are hard to attribute to concrete harness mechanisms. We introduce DemoEvolve, a demonstration-bootstrapped approach to harness evolution. When reward-only search is too broad and noisy, competent human trajectories serve as expert reference experience for the coding proposer, guiding harness-level diagnosis and editing. Experiments on Liar's Dice show that self-rollout evolution can work when episodes are short and failures are attributable. In contrast, Balatro exposes a harder long-horizon stochastic regime, where self-rollout evolution is misled by sparse feedback and candidate-selection noise, while tutorial-like textual knowledge alone does not yield stable improvement. Under the same limited budget, DemoEvolve produces more effective and auditable harness edits and achieves better performance. Overall, demonstrations make sparse-feedback harness evolution more diagnosable, localizable, and stable.

Yucheng Xin, Jiacheng Bao, Haoran Yang et al.

The integration of imitation and reinforcement learning has enabled remarkable advances in humanoid whole-body control, facilitating diverse human-like behaviors. However, research on environment-dependent motions remains limited. Existing methods typically enforce rigid trajectory tracking while neglecting physical interactions with the environment. We observe that humans naturally exploit a "weightless" state during non-self-stabilizing (NSS) motions--selectively relaxing specific joints to allow passive body--environment contact, thereby stabilizing the body and completing the motion. Inspired by this biological mechanism, we design a weightlessness-state auto-labeling strategy for dataset annotation; and we propose the Weightlessness Mechanism (WM), a method that dynamically determines which joints to relax and to what level, together enabling effective environmental interaction while executing target motions. We evaluate our approach on 3 representative NSS tasks: sitting on chairs of varying heights, lying down on beds with different inclinations, and leaning against walls via shoulder or elbow. Extensive experiments in simulation and on the Unitree G1 robot demonstrate that our WM method, trained on single-action demonstrations without any task-specific tuning, achieves strong generalization across diverse environmental configurations while maintaining motion stability. Our work bridges the gap between precise trajectory tracking and adaptive environmental interaction, offering a biologically-inspired solution for contact-rich humanoid control.

Yucheng Xin, Jiacheng Bao, Yubo Dong et al.

Humanoid robots have demonstrated impressive motor skills in a wide range of tasks, yet whole-body control for humanlike long-time, dynamic fighting remains particularly challenging due to the stringent requirements on agility and stability. While imitation learning enables robots to execute human-like fighting skills, existing approaches often rely on switching among multiple single-skill policies or employing a general policy to imitate input reference motions. These strategies suffer from instability when transitioning between skills, as the mismatch of initial and terminal states across skills or reference motions introduces out-of-domain disturbances, resulting in unsmooth or unstable behaviors. In this work, we propose RPG, a hybrid expert policy framework, for smooth and stable humanoid multi-skills transition. Our approach incorporates motion transition randomization and temporal randomization to train a unified policy that generates agile fighting actions with stability and smoothness during skill transitions. Furthermore, we design a control pipeline that integrates walking/running locomotion with fighting skills, allowing humanlike long-time combat of arbitrary duration that can be seamlessly interrupted or transit action policies at any time. Extensive experiments in simulation demonstrate the effectiveness of the proposed framework, and real-world deployment on the Unitree G1 humanoid robot further validates its robustness and applicability.

Yuhong Deng, Chongkun Xia, Xueqian Wang et al.

Rearranging deformable objects is a long-standing challenge in robotic manipulation for the high dimensionality of configuration space and the complex dynamics of deformable objects. We present a novel framework, Graph-Transporter, for goal-conditioned deformable object rearranging tasks. To tackle the challenge of complex configuration space and dynamics, we represent the configuration space of a deformable object with a graph structure and the graph features are encoded by a graph convolution network. Our framework adopts an architecture based on Fully Convolutional Network (FCN) to output pixel-wise pick-and-place actions from only visual input. Extensive experiments have been conducted to validate the effectiveness of the graph representation of deformable object configuration. The experimental results also demonstrate that our framework is effective and general in handling goal-conditioned deformable object rearranging tasks.

Lihui Xue, Zhihao Wang, Xueqian Wang et al.

Existing satellite remote sensing change detection (CD) methods often crop original large-scale bi-temporal image pairs into small patch pairs and then use pixel-level CD methods to fairly process all the patch pairs. However, due to the sparsity of change in large-scale satellite remote sensing images, existing pixel-level CD methods suffer from a waste of computational cost and memory resources on lots of unchanged areas, which reduces the processing efficiency of on-board platform with extremely limited computation and memory resources. To address this issue, we propose a lightweight patch-level CD network (LPCDNet) to rapidly remove lots of unchanged patch pairs in large-scale bi-temporal image pairs. This is helpful to accelerate the subsequent pixel-level CD processing stage and reduce its memory costs. In our LPCDNet, a sensitivity-guided channel pruning method is proposed to remove unimportant channels and construct the lightweight backbone network on basis of ResNet18 network. Then, the multi-layer feature compression (MLFC) module is designed to compress and fuse the multi-level feature information of bi-temporal image patch. The output of MLFC module is fed into the fully-connected decision network to generate the predicted binary label. Finally, a weighted cross-entropy loss is utilized in the training process of network to tackle the change/unchange class imbalance problem. Experiments on two CD datasets demonstrate that our LPCDNet achieves more than 1000 frames per second on an edge computation platform, i.e., NVIDIA Jetson AGX Orin, which is more than 3 times that of the existing methods without noticeable CD performance loss. In addition, our method reduces more than 60% memory costs of the subsequent pixel-level CD processing stage.

Haoyuan Sun, Bo Xia, Yongzhe Chang et al.

Direct Preference Optimization (DPO) has recently expanded its successful application from aligning large language models (LLMs) to aligning text-to-image models with human preferences, which has generated considerable interest within the community. However, we have observed that these approaches rely solely on minimizing the reverse Kullback-Leibler divergence during alignment process between the fine-tuned model and the reference model, neglecting the incorporation of other divergence constraints. In this study, we focus on extending reverse Kullback-Leibler divergence in the alignment paradigm of text-to-image models to $f$-divergence, which aims to garner better alignment performance as well as good generation diversity. We provide the generalized formula of the alignment paradigm under the $f$-divergence condition and thoroughly analyze the impact of different divergence constraints on alignment process from the perspective of gradient fields. We conduct comprehensive evaluation on image-text alignment performance, human value alignment performance and generation diversity performance under different divergence constraints, and the results indicate that alignment based on Jensen-Shannon divergence achieves the best trade-off among them. The option of divergence employed for aligning text-to-image models significantly impacts the trade-off between alignment performance (especially human value alignment) and generation diversity, which highlights the necessity of selecting an appropriate divergence for practical applications.

Shoujie Li, Changqing Guo, Jianle Xu et al.

Robotic manipulators for aerospace applications require a delicate balance between lightweight construction and fault-tolerant operation to satisfy strict weight limitations and ensure reliability in remote, hazardous environments. This paper presents Time-Division Multiplexing Actuation (TDMA), a practical approach for tendon-driven robots that significantly reduces actuator count while preserving high torque output and intrinsic fault tolerance. The key hardware employs a vertically-stacked rotational selection structure that integrates self-rotating TDM motors for rapid configuration, electromagnetic clutches enabling sub-0.1 second engagement, a worm gear reducer for enhanced load capacity and self-locking capability, and a dual-encoder system for precise, long-term positioning. Leveraging TDMA, the proposed MuxArm achieves a self-weight of 2.17 kg, supports an actuator driving capacity of 10 kg, and maintains end-effector accuracy up to 1% of its length, even under partial servo failure. Additionally, an actuation space trajectory planning algorithm is developed, enabling fault-tolerant control and reducing tendon load by up to 50% compared to conventional methods. Comprehensive experiments demonstrate MuxArm's robust performance in diverse settings, including free-space, cluttered, and confined environments.

Chenyang Cao, Yucheng Xin, Silang Wu et al.

Offline Safe Reinforcement Learning (RL) seeks to address safety constraints by learning from static datasets and restricting exploration. However, these approaches heavily rely on the dataset and struggle to generalize to unseen scenarios safely. In this paper, we aim to improve safety during the deployment of vision-based robotic tasks through online fine-tuning an offline pretrained policy. To facilitate effective fine-tuning, we introduce model-based RL, which is known for its data efficiency. Specifically, our method employs in-sample optimization to improve offline training efficiency while incorporating reachability guidance to ensure safety. After obtaining an offline safe policy, a safe policy expansion approach is leveraged for online fine-tuning. The performance of our method is validated on simulation benchmarks with five vision-only tasks and through real-world robot deployment using limited data. It demonstrates that our approach significantly improves the generalization of offline policies to unseen safety-constrained scenarios. To the best of our knowledge, this is the first work to explore offline-to-online RL for safe generalization tasks.

Haoyu Wang, Bingzhe Wu, Yatao Bian et al.

Large Language Models (LLMs) are implicit troublemakers. While they provide valuable insights and assist in problem-solving, they can also potentially serve as a resource for malicious activities. Implementing safety alignment could mitigate the risk of LLMs generating harmful responses. We argue that: even when an LLM appears to successfully block harmful queries, there may still be hidden vulnerabilities that could act as ticking time bombs. To identify these underlying weaknesses, we propose to use a cost value model as both a detector and an attacker. Trained on external or self-generated harmful datasets, the cost value model could successfully influence the original safe LLM to output toxic content in decoding process. For instance, LLaMA-2-chat 7B outputs 39.18% concrete toxic content, along with only 22.16% refusals without any harmful suffixes. These potential weaknesses can then be exploited via prompt optimization such as soft prompts on images. We name this decoding strategy: Jailbreak Value Decoding (JVD), emphasizing that seemingly secure LLMs may not be as safe as we initially believe. They could be used to gather harmful data or launch covert attacks.

Yuhong Deng, Xueqian Wang, Lipeng chen

Goal-conditioned rearrangement of deformable objects (e.g. straightening a rope and folding a cloth) is one of the most common deformable manipulation tasks, where the robot needs to rearrange a deformable object into a prescribed goal configuration with only visual observations. These tasks are typically confronted with two main challenges: the high dimensionality of deformable configuration space and the underlying complexity, nonlinearity and uncertainty inherent in deformable dynamics. To address these challenges, we propose a novel representation strategy that can efficiently model the deformable object states with a set of keypoints and their interactions. We further propose local-graph neural network (GNN), a light local GNN learning to jointly model the deformable rearrangement dynamics and infer the optimal manipulation actions (e.g. pick and place) by constructing and updating two dynamic graphs. Both simulated and real experiments have been conducted to demonstrate that the proposed dynamic graph representation shows superior expressiveness in modeling deformable rearrangement dynamics. Our method reaches much higher success rates on a variety of deformable rearrangement tasks (96.3% on average) than state-of-the-art method in simulation experiments. Besides, our method is much more lighter and has a 60% shorter inference time than state-of-the-art methods. We also demonstrate that our method performs well in the multi-task learning scenario and can be transferred to real-world applications with an average success rate of 95% by solely fine tuning a keypoint detector.

Shangyi Luo, Peng Sun, Ji Zhu et al.

With the increasing presence of service robots and autonomous vehicles in human environments, navigation systems need to evolve beyond simple destination reach to incorporate social awareness. This paper introduces GSON, a novel group-based social navigation framework that leverages Large Multimodal Models (LMMs) to enhance robots' social perception capabilities. Our approach uses visual prompting to enable zero-shot extraction of social relationships among pedestrians and integrates these results with robust pedestrian detection and tracking pipelines to overcome the inherent inference speed limitations of LMMs. The planning system incorporates a mid-level planner that sits between global path planning and local motion planning, effectively preserving both global context and reactive responsiveness while avoiding disruption of the predicted social group. We validate GSON through extensive real-world mobile robot navigation experiments involving complex social scenarios such as queuing, conversations, and photo sessions. Comparative results show that our system significantly outperforms existing navigation approaches in minimizing social perturbations while maintaining comparable performance on traditional navigation metrics.

Jiahua Pei, Yi Liu, Guoping Pan et al.

Object Goal Navigation (ObjectNav) refers to an agent navigating to an object in an unseen environment, which is an ability often required in the accomplishment of complex tasks. While existing methods demonstrate proficiency in isolated single object navigation, their limitations emerge in the restricted applicability of lifelong memory representations, which ultimately hinders effective navigation toward continual targets over extended periods. To address this problem, we propose OVAL, a novel lifelong open-vocabulary memory framework, which enables efficient and precise execution of long-term navigation in semantically open tasks. Within this framework, we introduce memory descriptors to facilitate structured management of the memory model. Additionally, we propose a novel probability-based exploration strategy, utilizing a multi-value frontier scoring to enhance lifelong exploration efficiency. Extensive experiments demonstrate the efficiency and robustness of the proposed system.

Tianyi Zeng, Jincheng Gao, Tianyi Wang et al.

Maintaining background consistency while enhancing foreground quality remains a core challenge in video editing. Injecting full-image information often leads to background artifacts, whereas rigid background locking severely constrains the model's capacity for foreground generation. To address this issue, we propose KV-Lock, a training-free framework tailored for DiT-based video diffusion models. Our core insight is that the hallucination metric (variance of denoising prediction) directly quantifies generation diversity, which is inherently linked to the classifier-free guidance (CFG) scale. Building upon this, KV-Lock leverages diffusion hallucination detection to dynamically schedule two key components: the fusion ratio between cached background key-values (KVs) and newly generated KVs, and the CFG scale. When hallucination risk is detected, KV-Lock strengthens background KV locking and simultaneously amplifies conditional guidance for foreground generation, thereby mitigating artifacts and improving generation fidelity. As a training-free, plug-and-play module, KV-Lock can be easily integrated into any pre-trained DiT-based models. Extensive experiments validate that our method outperforms existing approaches in improved foreground quality with high background fidelity across various video editing tasks.

Haoyuan Sun, Jiaqi Wu, Bo Xia et al.

Standing in 2025, at a critical juncture in the pursuit of Artificial General Intelligence (AGI), reinforcement fine-tuning (RFT) has demonstrated significant potential in enhancing the reasoning capability of large language models (LLMs) and has led to the development of cutting-edge AI models such as OpenAI-o1 and DeepSeek-R1. Moreover, the efficient application of RFT to enhance the reasoning capability of multimodal large language models (MLLMs) has attracted widespread attention from the community. In this position paper, we argue that reinforcement fine-tuning powers the reasoning capability of multimodal large language models. To begin with, we provide a detailed introduction to the fundamental background knowledge that researchers interested in this field should be familiar with. Furthermore, we meticulously summarize the improvements of RFT in powering reasoning capability of MLLMs into five key points: diverse modalities, diverse tasks and domains, better training algorithms, abundant benchmarks and thriving engineering frameworks. Finally, we propose five promising directions for future research that the community might consider. We hope that this position paper will provide valuable insights to the community at this pivotal stage in the advancement toward AGI. Summary of works done on RFT for MLLMs is available at https://github.com/Sun-Haoyuan23/Awesome-RL-based-Reasoning-MLLMs.

Chenyang Cao, Zichen Yan, Renhao Lu et al.

Offline goal-conditioned reinforcement learning (GCRL) aims at solving goal-reaching tasks with sparse rewards from an offline dataset. While prior work has demonstrated various approaches for agents to learn near-optimal policies, these methods encounter limitations when dealing with diverse constraints in complex environments, such as safety constraints. Some of these approaches prioritize goal attainment without considering safety, while others excessively focus on safety at the expense of training efficiency. In this paper, we study the problem of constrained offline GCRL and propose a new method called Recovery-based Supervised Learning (RbSL) to accomplish safety-critical tasks with various goals. To evaluate the method performance, we build a benchmark based on the robot-fetching environment with a randomly positioned obstacle and use expert or random policies to generate an offline dataset. We compare RbSL with three offline GCRL algorithms and one offline safe RL algorithm. As a result, our method outperforms the existing state-of-the-art methods to a large extent. Furthermore, we validate the practicality and effectiveness of RbSL by deploying it on a real Panda manipulator. Code is available at https://github.com/Sunlighted/RbSL.git.

Lirong Che, Zhenfeng Gan, Yanbo Chen et al.

Embodied agents for creative tasks like photography must bridge the semantic gap between high-level language commands and geometric control. We introduce PhotoAgent, an agent that achieves this by integrating Large Multimodal Models (LMMs) reasoning with a novel control paradigm. PhotoAgent first translates subjective aesthetic goals into solvable geometric constraints via LMM-driven, chain-of-thought (CoT) reasoning, allowing an analytical solver to compute a high-quality initial viewpoint. This initial pose is then iteratively refined through visual reflection within a photorealistic internal world model built with 3D Gaussian Splatting (3DGS). This ``mental simulation'' replaces costly and slow physical trial-and-error, enabling rapid convergence to aesthetically superior results. Evaluations confirm that PhotoAgent excels in spatial reasoning and achieves superior final image quality.

Qinglong Meng, Chongkun Xia, Xueqian Wang

Normalizing flow is a generative modeling approach with efficient sampling. However, Flow-based models suffer two issues: 1) If the target distribution is manifold, due to the unmatch between the dimensions of the latent target distribution and the data distribution, flow-based models might perform badly. 2) Discrete data might make flow-based models collapse into a degenerate mixture of point masses. To sidestep such two issues, we propose PaddingFlow, a novel dequantization method, which improves normalizing flows with padding-dimensional noise. To implement PaddingFlow, only the dimension of normalizing flows needs to be modified. Thus, our method is easy to implement and computationally cheap. Moreover, the padding-dimensional noise is only added to the padding dimension, which means PaddingFlow can dequantize without changing data distributions. Implementing existing dequantization methods needs to change data distributions, which might degrade performance. We validate our method on the main benchmarks of unconditional density estimation, including five tabular datasets and four image datasets for Variational Autoencoder (VAE) models, and the Inverse Kinematics (IK) experiments which are conditional density estimation. The results show that PaddingFlow can perform better in all experiments in this paper, which means PaddingFlow is widely suitable for various tasks. The code is available at: https://github.com/AdamQLMeng/PaddingFlow.

Yuxing Wang, Jie Li, Cong Yu et al.

The emergence of modular satellites marks a significant transformation in spacecraft engineering, introducing a new paradigm of flexibility, resilience, and scalability in space exploration endeavors. In addressing complex challenges such as attitude control, both the satellite's morphological architecture and the controller are crucial for optimizing performance. Despite substantial research on optimal control, there remains a significant gap in developing optimized and practical assembly strategies for modular satellites tailored to specific mission constraints. This research gap primarily arises from the inherently complex nature of co-optimizing design and control, a process known for its notorious bi-level optimization loop. Conventionally tackled through artificial evolution, this issue involves optimizing the morphology based on the fitness of individual controllers, which is sample-inefficient and computationally expensive. In this paper, we introduce a novel gradient-based approach to simultaneously optimize both morphology and control for modular satellites, enhancing their performance and efficiency in attitude control missions. Our Monte Carlo simulations demonstrate that this co-optimization approach results in modular satellites with better mission performance compared to those designed by evolution-based approaches. Furthermore, this study discusses potential avenues for future research.

Haoyuan Sun, Jing Wang, Yuxin Song et al.

Recently, post-training methods based on reinforcement learning, with a particular focus on Group Relative Policy Optimization (GRPO), have emerged as the robust paradigm for further advancement of text-to-image (T2I) models. However, these methods are often prone to reward hacking, wherein models exploit biases in imperfect reward functions rather than yielding genuine performance gains. In this work, we identify that normalization could lead to miscalibration and directly removing the prompt-level standard deviation term yields an optimal policy ascent direction that is linear in the advantage but still limits the separation of genuine signals from noise. To mitigate the above issues, we propose Super-Linear Advantage Shaping (SLAS) by revisiting the functional update from an information geometry perspective. By extending the Fisher-Rao information metric with advantage-dependent weighting, SLAS introduces a non-linear geometric structure that reshapes the local policy space. This design relaxes constraints along high-advantage directions to amplify informative updates, while tightening those in low-advantage regions to suppress illusory gradients. In addition, batch-level normalization is applied to stabilize training under varying reward scales. Extensive evaluations demonstrate that SLAS consistently surpasses the DanceGRPO baseline across multiple backbones and benchmarks. In particular, it yields faster training dynamics, improved out-of-domain performance on GenEval and UniGenBench++, and enhanced robustness to model scaling, while mitigating reward hacking and preserving semantic and compositional fidelity in generations.

Huayue Liang, Ruochong Li, Yaodong Yang et al.

High-quality teleoperated demonstrations are a primary bottleneck for imitation learning (IL) in dexterous manipulation. However, haptic feedback provides operators with real-time contact information, enabling real-time finger posture adjustments, and thereby improving demonstration quality. Existing dexterous teleoperation platforms typically omit haptic feedback and remain bulky and expensive. We introduce CDF-Glove, a lightweight and low cost cable-driven force-feedback glove. The real-time state is available for 20 finger degrees of freedom (DoF), of which 16 are directly sensed and 4 are passively coupled (inferred from kinematic constraints). We develop a kinematic model and control stack for the glove, and validate them across multiple robotic hands with diverse kinematics and DoF. The CDF-Glove achieves distal joint repeatability of 0.4 degrees, and delivers about 200 ms force feedback latency, yielding a 4x improvement in task success rate relative to no-feedback teleoperation. We collect two bimanual teleoperation datasets, on which we train and evaluate Diffusion Policy baselines. Compared to kinesthetic teaching, the policies trained in our teleoperated demonstrations increase the average success rate by 55% and reduce the mean completion time by approximately 15.2 seconds (a 47.2% relative reduction). In particular, the CDF-Glove costs approximately US$230. The code and designs are released as open source at https://cdfglove.github.io/.

Lirong Che, Shuo Wen, Shan Huang et al.

Real-world robotic tasks are long-horizon and often span multiple floors, demanding rich spatial reasoning. However, existing embodied benchmarks are largely confined to single-floor in-house environments, failing to reflect the complexity of real-world tasks. We introduce MANSION, the first language-driven framework for generating building-scale, multi-floor 3D environments. Being aware of vertical structural constraints, MANSION generates realistic, navigable whole-building structures with diverse, human-friendly scenes, enabling the development and evaluation of cross-floor long-horizon tasks. Building on this framework, we release MansionWorld, a dataset of over 1,000 diverse buildings ranging from hospitals to offices, alongside a Task-Semantic Scene Editing Agent that customizes these environments using open-vocabulary commands to meet specific user needs. Benchmarking reveals that state-of-the-art agents degrade sharply in our settings, establishing MANSION as a critical testbed for the next generation of spatial reasoning and planning.