论文检索

Muthukumar Pandaram, Jakob Hollenstein, David Drexel, Samuele Tosatto, Antonio Rodriguez-Sanchez, Justus Piater

The use of learned dynamics models, also known as world models, can improve the sample efficiency of reinforcement learning. Recent work suggests that the underlying causal graphs of such dynamics models are sparsely connected, with each of the future state variables depending only on a small subset of the current state variables, and that learning may therefore benefit from sparsity priors. Similarly, temporal sparsity, i.e. sparsely and abruptly changing local dynamics, has also been proposed as a useful inductive bias. In this work, we critically examine these assumptions by analyzing ground truth dynamics from a set of robotic reinforcement learning environments in the MuJoCo Playground benchmark suite, aiming to determine whether the proposed notions of state and temporal sparsity actually tend to hold in typical reinforcement learning tasks. We study (i) whether the causal graphs of environment dynamics are sparse, (ii) whether such sparsity is state-dependent, and (iii) whether local system dynamics change sparsely. Our results indicate that global sparsity is rare, but instead the tasks show local, state-dependent sparsity in their dynamics and this sparsity exhibits distinct structures, appearing in temporally localized clusters (e.g., during contact events) and affecting specific subsets of state dimensions. These findings challenge common sparsity prior assumptions in dynamics learning, emphasizing the need for grounded inductive biases that reflect the state-dependent sparsity structure of real-world dynamics.

Youngmin Oh, Hyung-Il Kim, Jung Uk Kim

Multi-task learning (MTL) is critical in real-world applications such as autonomous driving and robotics, enabling simultaneous handling of diverse tasks. However, obtaining fully annotated data for all tasks is impractical due to labeling costs. Existing methods for partially labeled MTL typically rely on predictions from unlabeled tasks, making it difficult to establish reliable task associations and potentially leading to negative transfer and suboptimal performance. To address these issues, we propose a prototype-based knowledge retrieval framework that achieves robust MTL instead of relying on predictions from unlabeled tasks. Our framework consists of two key components: (1) a task prototype embedding task-specific characteristics and quantifying task associations, and (2) a knowledge retrieval transformer that adaptively refines feature representations based on these associations. To achieve this, we introduce an association knowledge generating (AKG) loss to ensure the task prototype consistently captures task-specific characteristics. Extensive experiments demonstrate the effectiveness of our framework, highlighting its potential for robust multi-task learning, even when only a subset of tasks is annotated.

Asen Nachkov, Danda Pani Paudel, Jan-Nico Zaech, Davide Scaramuzza, Luc Van Gool

Differentiable simulators represent an environment’s dynamics as a differentiable function. Within robotics and autonomous driving, this property is used in Analytic Policy Gradients (APG), which relies on backpropagating through the dynamics to train accurate policies for diverse tasks. Here we show that differentiable simulation also has an important role in world modeling, where it can impart predictive, prescriptive, and counterfactual capabilities to an agent. Specifically, we design three novel task setups in which the differentiable dynamics are combined within an end-to-end computation graph not with a policy, but a state predictor. This allows us to learn relative odometry, optimal planners, and optimal inverse states. We collectively call these predictors Analytic World Models (AWMs) and demonstrate how differentiable simulation enables their efficient, end-to-end learning. In autonomous driving scenarios, they have broad applicability and can augment an agent’s decision-making beyond reactive control.

Xu Liu, Yu Liu, Hanshuo Qiu, Yang Qirong, Zhouhui Lian

Vision-Language Navigation (VLN) enables agents to navigate in complex environments by following natural language instructions grounded in visual observations. Although most existing work has focused on ground-based robots or outdoor Unmanned Aerial Vehicles (UAVs), indoor UAV-based VLN remains underexplored, despite its relevance to real-world applications such as inspection, delivery, and search-and-rescue in confined spaces. To bridge this gap, we introduce IndoorUAV, a novel benchmark and method specifically tailored for VLN with indoor UAVs. We begin by curating over 1,000 diverse and structurally rich 3D indoor scenes from the Habitat simulator. Within these environments, we simulate realistic UAV flight dynamics to collect diverse 3D navigation trajectories manually, further enriched through data augmentation techniques. Furthermore, we design an automated annotation pipeline to generate natural language instructions of varying granularity for each trajectory. This process yields over 16,000 high-quality trajectories, comprising the IndoorUAV-VLN subset, which focuses on long-horizon VLN. To support short-horizon planning, we segment long trajectories into sub-trajectories by selecting semantically salient keyframes and regenerating concise instructions, forming the IndoorUAV-VLA subset. Finally, we introduce IndoorUAV-Agent, a novel navigation model designed for our benchmark, leveraging task decomposition and multimodal reasoning. We hope IndoorUAV serves as a valuable resource to advance research on vision-language embodied AI in the indoor aerial navigation domain.

Bingyi Liu, Jinbo He, Haiyong Shi, Enshu Wang, Weizhen Han, Jingxiang Hao, Peixi Wang, Zhuangzhuang Zhang

Hybrid action space, which combines discrete choices and continuous parameters, is prevalent in domains such as robot control and game AI. However, efficiently modeling and optimizing hybrid discrete-continuous action space remains a fundamental challenge, mainly due to limited policy expressiveness and poor scalability in high-dimensional settings. To address this challenge, we view the hybrid action space problem as a fully cooperative game and propose a Cooperative Hybrid Diffusion Policies (CHDP) framework to solve it. CHDP employs two cooperative agents that leverage a discrete and a continuous diffusion policy, respectively. The continuous policy is conditioned on the discrete action's representation, explicitly modeling the dependency between them. This cooperative design allows the diffusion policies to leverage their expressiveness to capture complex distributions in their respective action spaces. To mitigate the update conflicts arising from simultaneous policy updates in this cooperative setting, we employ a sequential update scheme that fosters co-adaptation. Moreover, to improve scalability when learning in high-dimensional discrete action space, we construct a codebook that embeds the action space into a low-dimensional latent space. This mapping enables the discrete policy to learn in a compact, structured space. Finally, we design a Q-function-based guidance mechanism to align the codebook's embeddings with the discrete policy's representation during training. On challenging hybrid action benchmarks, CHDP outperforms state-of-the-art method by up to 19.3% in success rate.

Xun Liang, Xin Guo, Zhongming Jin, Weihang Pan, Penghui Shang, Deng Cai, Binbin Lin, Jieping Ye

The spatial reasoning task aims to reason about the spatial relationships in 2D and 3D space, which is a fundamental capability for Visual Question Answering (VQA) and robotics. Although vision language models (VLMs) have developed rapidly in recent years, they are still struggling with the spatial reasoning task. In this paper, we introduce a method that can enhance Spatial reasoning through Visual and Textual thinking Simultaneously (SpatialVTS). In the spatial visual thinking phase, our model is trained to generate location-related specific tokens of important targets automatically. Not only are the objects mentioned in the problem addressed, but also the potential objects related to the reasoning are considered. During the spatial textual thinking phase, our model conducts long-term thinking based on visual cues and dialogues and gradually inferences the answers to spatial reasoning problems. To effectively support the model's training, we made manual corrections to the existing spatial reasoning dataset, eliminating numerous incorrect labels resulting from automatic annotation, restructuring the data input format to enhance generalization, and developing a reasoning framework for model thinking. Without introducing any additional information (such as masks or depth), our model's overall average level in several spatial understanding tasks has significantly improved compared with other models.

Jinhao Liang, Sven Koenig, Ferdinando Fioretto

Multi-Robot Motion Planning (MRMP) involves generating collision-free trajectories for multiple robots operating in a shared continuous workspace. While discrete multi-agent path finding (MAPF) methods are broadly adopted due to their scalability, their coarse discretization severely limits trajectory quality. In contrast, continuous optimization-based planners offer higher-quality paths but suffer from the curse of dimensionality, resulting in poor scalability with respect to the number of robots. This paper tackles the limitations of these two approaches by introducing a novel framework that integrates discrete MAPF solvers with constrained generative diffusion models. The resulting framework, called Discrete-Guided Diffusion (DGD), has three key characteristics: (1) it decomposes the original nonconvex MRMP problem into tractable subproblems with convex configuration spaces, (2) it combines discrete MAPF solutions with constrained optimization techniques to guide diffusion models capture complex spatiotemporal dependencies among robots, and (3) it incorporates a lightweight constraint repair mechanism to ensure trajectory feasibility. The proposed method sets a new state-of-the-art performance in large-scale, complex environments, scaling to 100 robots while achieving planning efficiency and high success rates.

Yuxuan Li, Yicheng Gao, Ning Yang, Stephen Xia

Episodic tasks in Reinforcement Learning (RL) often pose challenges due to sparse reward signals and high-dimensional state spaces, which hinder efficient learning. Additionally, these tasks often feature hidden “trap states”—irreversible failures that prevent task completion but do not provide explicit negative rewards to guide agents away from repeated errors. To address these issues, we propose Time-Weighted Contrastive Reward Learning (TW-CRL), an Inverse Reinforcement Learning (IRL) framework that leverages both successful and failed demonstrations. By incorporating temporal information, TW-CRL learns a dense reward function that identifies critical states associated with success or failure. This approach not only enables agents to avoid trap states but also encourages meaningful exploration beyond simple imitation of expert trajectories. Empirical evaluations on navigation tasks and robotic manipulation benchmarks demonstrate that TW-CRL surpasses state-of-the-art methods, achieving improved efficiency and robustness.

Tianyu Guo, Shanwei Zhao, Shiai Zhu, Chenguang Ma

Deploying Vision-Language Models (VLMs) on edge devices (e.g., smartphones and robots) is crucial for enabling low-latency and privacy-preserving intelligent applications. Given the resource constraints of these devices, quantization offers a promising solution by improving memory efficiency and reducing bandwidth requirements, thereby facilitating the deployment of VLMs. However, existing research has rarely explored aggressive quantization on VLMs, particularly for the models ranging from 1B to 2B parameters, which are more suitable for resource-constrained edge devices. In this paper, we propose SPEED-Q, a novel Staged Processing with EnhancEd Distillation framework for VLM low-bit weight-only quantization that systematically addresses the following two critical obstacles: (1) significant discrepancies in quantization sensitivity between vision (ViT) and language (LLM) components in VLMs; (2) training instability arising from the reduced numerical precision inherent in low-bit quantization. In SPEED-Q, a staged sensitivity adaptive mechanism is introduced to effectively harmonize performance across different modalities. We further propose a distillation-enhanced quantization strategy to stabilize the training process and reduce data dependence. Together, SPEED-Q enables accurate, stable, and data-efficient quantization of complex VLMs. SPEED-Q is the first framework tailored for quantizing entire small-scale billion-parameter VLMs to low bits. Extensive experiments across multiple benchmarks demonstrate that SPEED-Q achieves up to 6x higher accuracy than existing quantization methods under 2-bit settings and consistently outperforms prior on-device VLMs under both 2-bit and 4-bit settings.

Bram Grooten, Patrick MacAlpine, Kaushik Subramanian, Peter Stone, Peter R. Wurman

Generalization to unseen environments is a significant challenge in the field of robotics and control. In this work, we focus on contextual reinforcement learning, where agents act within environments with varying contexts, such as self-driving cars or quadrupedal robots that need to operate in different terrains or weather conditions than they were trained for. We tackle the critical task of generalizing to out-of-distribution (OOD) settings, without access to explicit context information at test time. Recent work has addressed this problem by training a context encoder and a history adaptation module in separate stages. While promising, this two-phase approach is cumbersome to implement and train. We simplify the methodology and introduce SPARC: single-phase adaptation for robust control. We test SPARC on varying contexts within the high-fidelity racing simulator Gran Turismo 7 and wind-perturbed MuJoCo environments, and find that it achieves reliable and robust OOD generalization.

Yonghao Dong, Qiang He, Penghong Rui, Zhenzhe Zheng, Zhao Li, Feifei Chen, Hai Jin, Yun Yang

Vision Transformers (ViTs) have gained significant attention and widespread adoption due to their impressive performance in various computer vision tasks. However, in practice, their substantial computational overhead often leads to high inference latency and increased overheads when deployed on resource-constrained edge devices like smartphones, autonomous vehicles, and robots. To address these challenges, Early Exit (EE) has emerged as a promising approach for lightweight inference on edge devices. It accelerates inference and reduces computational overhead by adaptively producing predictions through early exits based on sample complexity. Existing EE methods typically suffer from substantial accuracy decreases in late exits while providing only marginal accuracy improvements to early exits. This paper presents EnViT, an exit-aware structured dropout-enabled self-distillation approach that enhances the performance of early exits without compromising late exits. EnViT leverages structured dropout to enable self-distillation, where the full model serves as the teacher and its own virtual sub-models generated by structured dropout as students. This mechanism effectively distills knowledge from the full model to early exits and avoids performance degradation in late exits by mitigating parameter conflicts across exits during training. Evaluation on five datasets shows that our EnViT achieves accuracy improvements ranging from 0.36% to 7.92% while maintaining competitive speed-up ratios of 1.72x to 2.23x.

Jingxiao Chen, Weiji Xie, Weinan Zhang, Yong Yu, Ying Wen

Offline Reinforcement Learning (RL) enables policy improvement from fixed datasets without online interactions, making it highly suitable for real-world applications lacking efficient simulators. Despite its success in the single-agent setting, offline multi-agent RL remains a challenge, especially in competitive games. Firstly, unaware of the game structure, it is impossible to interact with the opponents and conduct a major learning paradigm, self-play, for competitive games. Secondly, real-world datasets cannot cover all the state and action space in the game, resulting in barriers to identifying Nash equilibrium (NE). To address these issues, this paper introduces Off-FSP, the first practical model-free offline RL algorithm for competitive games. We start by simulating interactions with various opponents by adjusting the weights of the fixed dataset with importance sampling. This technique allows us to learn the best responses to different opponents and employ the Offline Self-Play learning framework. To overcome the challenge of partial coverage, we combine the single-agent offline RL method with Fictitious Self-Play (FSP) to approximate NE by constraining the approximate best responses away from out-of-distribution actions. Experiments on matrix games, extensive-form poker, and board games demonstrate that Off-FSP achieves significantly lower exploitability than state-of-the-art baselines. Finally, we validate Off-FSP on a real-world human-robot competitive task, demonstrating its potential for solving complex, hard-to-simulate real-world problems.

Jianlei Chang, Ruofeng Mei, Wei Ke, Xiangyu Xu

Generative modeling has recently shown remarkable promise for visuomotor policy learning, enabling flexible and expressive control across diverse embodied AI tasks. However, existing generative policies often struggle with data inefficiency, requiring large-scale demonstrations, and sampling inefficiency, incurring slow action generation during inference. We introduce EfficientFlow, a unified framework for efficient embodied AI with flow-based policy learning. To enhance data efficiency, we bring equivariance into flow matching. We theoretically prove that when using an isotropic Gaussian prior and an equivariant velocity prediction network, the resulting action distribution remains equivariant, leading to improved generalization and substantially reduced data demands. To accelerate sampling, we propose a novel acceleration regularization strategy. As direct computation of acceleration is intractable for marginal flow trajectories, we derive a novel surrogate loss that enables stable and scalable training using only conditional trajectories. Across a wide range of robotic manipulation benchmarks, the proposed algorithm achieves competitive or superior performance under limited data while offering dramatically faster inference. These results highlight EfficientFlow as a powerful and efficient paradigm for high-performance embodied AI.

Guowei Zou, Weibing Li, Hejun Wu, Yukun Qian, Yuhang Wang, Haitao Wang

Diffusion policies excel at robotic manipulation by naturally modeling multimodal action distributions in high-dimensional spaces. Nevertheless, diffusion policies suffer from diffusion representation collapse: semantically similar observations are mapped to indistinguishable features, ultimately impairing their ability to handle subtle but critical variations required for complex robotic manipulation. To address this problem, we propose D²PPO (Diffusion Policy Policy Optimization with Dispersive Loss). D²PPO introduces dispersive loss regularization that combats representation collapse by treating all hidden representations within each batch as negative pairs. D²PPO compels the network to learn discriminative representations of similar observations, thereby enabling the policy to identify subtle yet crucial differences necessary for precise manipulation. In evaluation, we find that early-layer regularization benefits simple tasks, while late-layer regularization sharply enhances performance on complex manipulation tasks. On RoboMimic benchmarks, D²PPO achieves an average improvement of 22.7% in pre-training and 26.1% after fine-tuning, setting new SOTA results. In comparison with SOTA, the results of real-world experiments on a Franka Emika Panda robot show the excitingly high success rate of our method. The superiority of our method is especially evident in complex tasks.

Yijie Zhu, Rui Shao, Ziyang Liu, Jie He, Jizhihui Liu, Jiuru Wang, Zitong Yu

Unified video and action prediction models hold great potential for robotic manipulation, as future observations offer contextual cues for planning, while actions reveal how interactions shape the environment. However, most existing approaches treat observation and action generation in a monolithic and goal-agnostic manner, often leading to semantically misaligned predictions and incoherent behaviors. To this end, we propose H-GAR, a Hierarchical interaction framework via Goal-driven observation-Action Refinement. To anchor prediction to the task objective, H-GAR first produces a goal observation and a coarse action sketch that outline a high-level route toward the goal. To enable explicit interaction between observation and action under the guidance of the goal observation for more coherent decision-making, we devise two synergistic modules. (1) Goal-Conditioned Observation Synthesizer (GOS) synthesizes intermediate observations based on the coarse-grained actions and the predicted goal observation. (2) Interaction-Aware Action Refiner (IAAR) refines coarse actions into fine-grained, goal-consistent actions by leveraging feedback from the intermediate observations and a Historical Action Memory Bank that encodes prior actions to ensure temporal consistency. By integrating goal grounding with explicit action-observation interaction in a coarse-to-fine manner, H-GAR enables more accurate manipulation. Extensive experiments on both simulation and real-world robotic manipulation tasks demonstrate that H-GAR achieves state-of-the-art performance.

Xingyu Zhu, Zhiwen Tu, Yan Wu, Shan Luo, Hechang Chen, Yixing Gao

Enabling robots to grasp disorganized cloth for efficient storage is valuable in robot-assisted room organization. Diverse deformations of cloth and the stacking of multiple items limit grasping-pose estimation that relies on annotations. This necessitates segmenting each cloth item in an unsupervised manner before estimating the grasping position. However, existing segmentation methods primarily focus on improving metrics such as Intersection-over-Union and Pixel Accuracy, which cannot effectively measure the segmentation errors of the cloth area and thus lead to failure grasping position estimation. To address this challenge, we use False Discovery Rate (FDR) as a novel measure of segmentation errors and analyze its impact on grasping success. Our preliminary study reveals a negative correlation between segmentation FDR and grasping success rate, highlighting the need for more reliable segmentation in cluttered cloth scenarios. Therefore, we propose an unsupervised cloth segmentation network based on feature distance-weighted constraints, designed to reduce the false discovery rate in cloth area perception without requiring expensive pixel-level manual annotations. Additionally, to estimate the grasping position on the perceived cloth area, we introduce a strategy based on cloth surface wrinkle analysis, which operates without the need for annotations or training. By integrating the proposed segmentation network and grasping strategy, we develop a robotic system capable of sequentially grasping cluttered cloth from a table. Extensive real-world robotic experiments demonstrate the effectiveness of our approach, outperforming multiple baseline methods in segmentation FDR and grasping success rate.

Yiyun Zhou, Mingjing Xu, Jingwei Shi, Quanjiang Li, Jingyuan Chen

Tactile sensing offers rich and complementary information to vision and language, enabling robots to perceive fine-grained object properties. However, existing tactile sensors lack standardization, leading to redundant features that hinder cross-sensor generalization. Moreover, existing methods fail to fully integrate the intermediate communication among tactile, language, and vision modalities. To address this, we propose TLV-CoRe, a CLIP-based Tactile-Language-Vision Collaborative Representation learning method. TLV-CoRe introduces a Sensor-Aware Modulator to unify tactile features across different sensors and employs tactile-irrelevant decoupled learning to disentangle irrelevant tactile features. Additionally, a Unified Bridging Adapter is introduced to enhance tri-modal interaction within the shared representation space. To fairly evaluate the effectiveness of tactile models, we further propose the RSS evaluation framework, focusing on Robustness, Synergy, and Stability across different methods. Experimental results demonstrate that TLV-CoRe significantly improves sensor-agnostic representation learning and cross-modal alignment, offering a new direction for multimodal tactile representation.

Yifan Zhong, Xuchuan Huang, Ruochong Li, Ceyao Zhang, Zhang Chen, Tianrui Guan, Fanlian Zeng, Ka Nam Lui, Yuyao Ye, Yitao Liang 等

Dexterous grasping remains a fundamental yet challenging problem in robotics. A general-purpose robot must be capable of grasping diverse objects in arbitrary scenarios. However, existing research typically relies on restrictive assumptions, such as single-object settings or limited environments, showing constrained generalization. We present DexGraspVLA, a hierarchical framework for robust generalization in language-guided general dexterous grasping and beyond. It utilizes a pre-trained Vision-Language model as the high-level planner and learns a diffusion-based low-level Action controller. The key insight to achieve generalization lies in iteratively transforming diverse language and visual inputs into domain-invariant representations via foundation models, where imitation learning can be effectively applied due to the alleviation of domain shift. Notably, our method achieves a 90+% dexterous grasping success rate under thousands of challenging unseen cluttered scenes. Empirical analysis confirms the consistency of internal model behavior across environmental variations, validating our design. DexGraspVLA also, for the first time, simultaneously demonstrates free-form long-horizon prompt execution, robustness to adversarial objects and human disturbance, and failure recovery. Extended application to nonprehensile grasping further proves its generality.

Yingnan Zhao, Xinmiao Wang, Dewei Wang, Xinzhe Liu, Dan Lu, Qilong Han, Peng Liu, Chenjia Bai

Humanoid robots are promising to learn a diverse set of human-like locomotion behaviors, including standing up, walking, running, and jumping. However, existing methods predominantly require training independent policies for each skill, yielding behavior-specific controllers that exhibit limited generalization and brittle performance when deployed on irregular terrains and in diverse situations. To address this challenge, we propose Adaptive Humanoid Control (AHC) that adopts a two-stage framework to learn an adaptive humanoid locomotion controller across different skills and terrains. Specifically, we first train several primary locomotion policies and perform a multi-behavior distillation process to obtain a basic multi-behavior controller, facilitating adaptive behavior switching based on the environment. Then, we perform reinforced fine-tuning by collecting online feedback in performing adaptive behaviors on more diverse terrains, enhancing terrain adaptability for the adaptive behavior controller. We conduct experiments in both simulation and real-world experiments in Unitree G1 robots. The results show that our method exhibits strong adaptability across various situations and terrains.

Yuchi Zhang, Churui Sun, Shiqi Liang, Diyuan Liu, Chao Ji, Weinan Zhang, Ting Liu

Recent end-to-end robotic manipulation research increasingly adopts architectures inspired by large language models to enable robust manipulation. However, a critical challenge arises from severe distribution shifts between robotic action data, primarily due to substantial numerical variations in action commands across diverse robotic platforms and tasks, hindering the effective transfer of pretrained knowledge. To address this limitation, we propose a semantically grounded linguistic representation to normalize actions for efficient pretraining. Unlike conventional discretized action representations that are sensitive to numerical scales, the motion representation specifically disregards numeric scale effects, emphasizing directionality instead. This abstraction mitigates distribution shifts, yielding a more generalizable pretraining representation. Moreover, using the motion representation narrows the feature distance between action tokens and standard vocabulary tokens, mitigating modality gaps. Multi-task experiments on two benchmarks demonstrate that the proposed method significantly improves generalization performance and transferability in robotic manipulation tasks.