Vision-Language-Action (VLA) models are emerging as a next-generation paradigm for robotics. We introduce dVLA, a diffusion-based VLA that leverages a multimodal chain-of-thought to unify visual perception, language reasoning, and robotic control in a single system. dVLA jointly optimizes perception, language understanding, and action under a single diffusion objective, enabling stronger cross-modal reasoning and better generalization to novel instructions and objects. For practical deployment, we mitigate inference latency by incorporating two acceleration strategies, a prefix attention mask and KV caching, yielding up to around times speedup at test-time inference. We evaluate dVLA in both simulation and the real world: on the LIBERO benchmark, it achieves state-of-the-art performance with a 96.4% average success rate, consistently surpassing both discrete and continuous action policies; on a real Franka robot, it succeeds across a diverse task suite, including a challenging bin-picking task that requires multi-step planning, demonstrating robust real-world performance. Together, these results underscore the promise of unified diffusion frameworks for practical, high-performance VLA robotics.

We present ActiveUMI, a framework for a data collection system that transfers in-the-wild human demonstrations to robots capable of complex bimanual manipulation. ActiveUMI couples a portable VR teleoperation kit with sensorized controllers that mirror the robot's end-effectors, bridging human-robot kinematics via precise pose alignment. To ensure mobility and data quality, we introduce several key techniques, including immersive 3D model rendering, a self-contained wearable computer, and efficient calibration methods. ActiveUMI's defining feature is its capture of active, egocentric perception. By recording an operator's deliberate head movements via a head-mounted display, our system learns the crucial link between visual attention and manipulation. We evaluate ActiveUMI on six challenging bimanual tasks. Policies trained exclusively on ActiveUMI data achieve an average success rate of 70\% on in-distribution tasks and demonstrate strong generalization, retaining a 56\% success rate when tested on novel objects and in new environments. Our results demonstrate that portable data collection systems, when coupled with learned active perception, provide an effective and scalable pathway toward creating generalizable and highly capable real-world robot policies.

A significant bottleneck in humanoid policy learning is the acquisition of large-scale, diverse datasets, as collecting reliable real-world data remains both difficult and cost-prohibitive. To address this limitation, we introduce HumanoidExo, a novel system that transfers human motion to whole-body humanoid data. HumanoidExo offers a high-efficiency solution that minimizes the embodiment gap between the human demonstrator and the robot, thereby tackling the scarcity of whole-body humanoid data. By facilitating the collection of more voluminous and diverse datasets, our approach significantly enhances the performance of humanoid robots in dynamic, real-world scenarios. We evaluated our method across three challenging real-world tasks: table-top manipulation, manipulation integrated with stand-squat motions, and whole-body manipulation. Our results empirically demonstrate that HumanoidExo is a crucial addition to real-robot data, as it enables the humanoid policy to generalize to novel environments, learn complex whole-body control from only five real-robot demonstrations, and even acquire new skills (i.e., walking) solely from HumanoidExo data.

Vision-language-action (VLA) models have emerged as the next generation of models in robotics. However, despite leveraging powerful pre-trained Vision-Language Models (VLMs), existing end-to-end VLA systems often lose key capabilities during fine-tuning as the model adapts to specific robotic tasks. We argue that a generalizable VLA model should retain and expand upon the VLM's core competencies: 1) Open-world embodied reasoning - the VLA should inherit the knowledge from VLM, i.e., recognize anything that the VLM can recognize, be capable of solving math problems, and possess visual-spatial intelligence, 2) Reasoning following - effectively translating the open-world reasoning into actionable steps for the robot. In this work, we introduce ChatVLA-2, a novel mixture-of-expert VLA model coupled with a specialized two-stage training pipeline designed to preserve the VLM's original strengths while enabling actionable reasoning. To validate our approach, we design a math-matching task wherein a robot interprets math problems written on a whiteboard and picks corresponding number cards from a table to solve equations. Remarkably, our method exhibits exceptional mathematical reasoning and OCR capabilities, despite these abilities not being explicitly trained within the VLA. Furthermore, we demonstrate that the VLA possesses strong spatial reasoning skills, enabling it to interpret novel directional instructions involving previously unseen objects. Overall, our method showcases reasoning and comprehension abilities that significantly surpass state-of-the-art imitation learning methods such as OpenVLA, DexVLA, and pi-zero. This work represents a substantial advancement toward developing truly generalizable robotic foundation models endowed with robust reasoning capacities.

The field of robotics has made significant strides toward developing generalist robot manipulation policies. However, evaluating these policies in real-world scenarios remains time-consuming and challenging, particularly as the number of tasks scales and environmental conditions change. In this work, we demonstrate that world models can serve as a scalable, reproducible, and reliable proxy for real-world robot policy evaluation. A key challenge is generating accurate policy videos from world models that faithfully reflect the robot actions. We observe that directly inputting robot actions or using high-dimensional encoding methods often fails to generate action-following videos. To address this, we propose Policy2Vec, a simple yet effective approach to turn a video generation model into a world simulator that follows latent action to generate the robot video. We then introduce WorldEval, an automated pipeline designed to evaluate real-world robot policies entirely online. WorldEval effectively ranks various robot policies and individual checkpoints within a single policy, and functions as a safety detector to prevent dangerous actions by newly developed robot models. Through comprehensive paired evaluations of manipulation policies in real-world environments, we demonstrate a strong correlation between policy performance in WorldEval and real-world scenarios. Furthermore, our method significantly outperforms popular methods such as real-to-sim approach.

Humans possess a unified cognitive ability to perceive, comprehend, and interact with the physical world. Why can't large language models replicate this holistic understanding? Through a systematic analysis of existing training paradigms in vision-language-action models (VLA), we identify two key challenges: spurious forgetting, where robot training overwrites crucial visual-text alignments, and task interference, where competing control and understanding tasks degrade performance when trained jointly. To overcome these limitations, we propose ChatVLA, a novel framework featuring Phased Alignment Training, which incrementally integrates multimodal data after initial control mastery, and a Mixture-of-Experts architecture to minimize task interference. ChatVLA demonstrates competitive performance on visual question-answering datasets and significantly surpasses state-of-the-art vision-language-action (VLA) methods on multimodal understanding benchmarks. Notably, it achieves a six times higher performance on MMMU and scores 47.2% on MMStar with a more parameter-efficient design than ECoT. Furthermore, ChatVLA demonstrates superior performance on 25 real-world robot manipulation tasks compared to existing VLA methods like OpenVLA. Our findings highlight the potential of our unified framework for achieving both robust multimodal understanding and effective robot control.

Enabling robots to perform diverse tasks across varied environments is a central challenge in robot learning. While vision-language-action (VLA) models have shown promise for generalizable robot skills, realizing their full potential requires addressing limitations in action representation and efficient training. Current VLA models often focus on scaling the vision-language model (VLM) component, while the action space representation remains a critical bottleneck. This paper introduces DexVLA, a novel framework designed to enhance the efficiency and generalization capabilities of VLAs for complex, long-horizon tasks across diverse robot embodiments. DexVLA features a novel diffusion-based action expert, scaled to one billion parameters, designed for cross-embodiment learning. A novel embodiment curriculum learning strategy facilitates efficient training: (1) pre-training the diffusion expert that is separable from the VLA on cross-embodiment data, (2) aligning the VLA model to specific embodiments, and (3) post-training for rapid adaptation to new tasks. We conduct comprehensive experiments across multiple embodiments, including single-arm, bimanual, and dexterous hand, demonstrating DexVLA's adaptability to challenging tasks without task-specific adaptation, its ability to learn dexterous skills on novel embodiments with limited data, and its capacity to complete complex, long-horizon tasks using only direct language prompting, such as laundry folding. In all settings, our method demonstrates superior performance compared to state-of-the-art models like Octo, OpenVLA, and Diffusion Policy.

Robot foundation models, particularly Vision-Language-Action (VLA) models, have garnered significant attention for their ability to enhance robot policy learning, greatly improving robot generalization and robustness. OpenAI recent model, o1, showcased impressive capabilities in solving complex problems by utilizing extensive reasoning chains. This prompts an important question: can robot models achieve better performance in multi-task, complex environments by reviewing prior observations and then providing task-specific reasoning to guide action prediction? In this paper, we introduce Chain-of-Affordance (CoA), a novel approach to scaling robot models by incorporating reasoning in the format of sequential robot affordances to facilitate task completion. Specifically, we prompt the model to consider the following four types of affordances before taking action: a) object affordance - what object to manipulate and where it is; b) grasp affordance - the specific object part to grasp; c) spatial affordance - the optimal space to place the object; and d) movement affordance - the collision-free path for movement. By integrating this knowledge into the policy model, the robot gains essential context, allowing it to act with increased precision and robustness during inference. Our experiments demonstrate that CoA achieves superior performance than state-of-the-art robot foundation models, such as OpenVLA and Octo. Additionally, CoA shows strong generalization to unseen object poses, identifies free space, and avoids obstacles in novel environments.

Vision-Language-Action (VLA) models excel at robotic tasks by leveraging large-scale 2D vision-language pretraining, but their reliance on RGB images limits spatial reasoning critical for real-world interaction. Retraining these models with 3D data is computationally prohibitive, while discarding existing 2D datasets wastes valuable resources. To bridge this gap, we propose PointVLA, a framework that enhances pre-trained VLAs with point cloud inputs without requiring retraining. Our method freezes the vanilla action expert and injects 3D features via a lightweight modular block. To identify the most effective way of integrating point cloud representations, we conduct a skip-block analysis to pinpoint less useful blocks in the vanilla action expert, ensuring that 3D features are injected only into these blocks--minimizing disruption to pre-trained representations. Extensive experiments demonstrate that PointVLA outperforms state-of-the-art 2D imitation learning methods, such as OpenVLA, Diffusion Policy and DexVLA, across both simulated and real-world robotic tasks. Specifically, we highlight several key advantages of PointVLA enabled by point cloud integration: (1) Few-shot multi-tasking, where PointVLA successfully performs four different tasks using only 20 demonstrations each; (2) Real-vs-photo discrimination, where PointVLA distinguishes real objects from their images, leveraging 3D world knowledge to improve safety and reliability; (3) Height adaptability, Unlike conventional 2D imitation learning methods, PointVLA enables robots to adapt to objects at varying table height that unseen in train data. Furthermore, PointVLA achieves strong performance in long-horizon tasks, such as picking and packing objects from a moving conveyor belt, showcasing its ability to generalize across complex, dynamic environments.

In this paper, we present DiffusionVLA, a novel framework that seamlessly combines the autoregression model with the diffusion model for learning visuomotor policy. Central to our approach is a next-token prediction objective, enabling the model to reason effectively over the user's query in the context of current observations. Subsequently, a diffusion model is attached to generate robust action outputs. To enhance policy learning through self-reasoning, we introduce a novel reasoning injection module that integrates reasoning phrases directly into the policy learning process. The whole framework is simple and flexible, making it easy to deploy and upgrade. We conduct extensive experiments using multiple real robots to validate the effectiveness of DiffusionVLA. Our tests include a challenging factory sorting task, where DiffusionVLA successfully categorizes objects, including those not seen during training. We observe that the reasoning module makes the model interpretable. It allows observers to understand the model thought process and identify potential causes of policy failures. Additionally, we test DiffusionVLA on a zero-shot bin-picking task, achieving 63.7% accuracy on 102 previously unseen objects. Our method demonstrates robustness to visual changes, such as distractors and new backgrounds, and easily adapts to new embodiments. Furthermore, DiffusionVLA can follow novel instructions and retain conversational ability. Notably, DiffusionVLA is data-efficient and fast at inference; our smallest DiffusionVLA-2B runs 82Hz on a single A6000 GPU and can train from scratch on less than 50 demonstrations for a complex task. Finally, we scale the model from 2B to 72B parameters, showcasing improved generalization capabilities with increased model size.

Vision-Language-Action (VLA) models have shown remarkable potential in visuomotor control and instruction comprehension through end-to-end learning processes. However, current VLA models face significant challenges: they are slow during inference and require extensive pre-training on large amounts of robotic data, making real-world deployment difficult. In this paper, we introduce a new family of compact vision-language-action models, called TinyVLA, which offers two key advantages over existing VLA models: (1) faster inference speeds, and (2) improved data efficiency, eliminating the need for pre-training stage. Our framework incorporates two essential components to build TinyVLA: (1) initializing the policy backbone with robust, high-speed multimodal models, and (2) integrating a diffusion policy decoder during fine-tuning to enable precise robot actions. We conducted extensive evaluations of TinyVLA in both simulation and on real robots, demonstrating that our approach significantly outperforms the state-of-the-art VLA model, OpenVLA, in terms of speed and data efficiency, while delivering comparable or superior performance. Additionally, TinyVLA exhibits strong generalization capabilities across various dimensions, including language instructions, novel objects, unseen positions, changes in object appearance, background variations, and environmental shifts, often matching or exceeding the performance of OpenVLA. We believe that \methodname offers an interesting perspective on utilizing pre-trained multimodal models for policy learning.

Learning visuomotor policy for multi-task robotic manipulation has been a long-standing challenge for the robotics community. The difficulty lies in the diversity of action space: typically, a goal can be accomplished in multiple ways, resulting in a multimodal action distribution for a single task. The complexity of action distribution escalates as the number of tasks increases. In this work, we propose Discrete Policy, a robot learning method for training universal agents capable of multi-task manipulation skills. Discrete Policy employs vector quantization to map action sequences into a discrete latent space, facilitating the learning of task-specific codes. These codes are then reconstructed into the action space conditioned on observations and language instruction. We evaluate our method on both simulation and multiple real-world embodiments, including both single-arm and bimanual robot settings. We demonstrate that our proposed Discrete Policy outperforms a well-established Diffusion Policy baseline and many state-of-the-art approaches, including ACT, Octo, and OpenVLA. For example, in a real-world multi-task training setting with five tasks, Discrete Policy achieves an average success rate that is 26% higher than Diffusion Policy and 15% higher than OpenVLA. As the number of tasks increases to 12, the performance gap between Discrete Policy and Diffusion Policy widens to 32.5%, further showcasing the advantages of our approach. Our work empirically demonstrates that learning multi-task policies within the latent space is a vital step toward achieving general-purpose agents.

Diffusion Policy is a powerful technique tool for learning end-to-end visuomotor robot control. It is expected that Diffusion Policy possesses scalability, a key attribute for deep neural networks, typically suggesting that increasing model size would lead to enhanced performance. However, our observations indicate that Diffusion Policy in transformer architecture (\DP) struggles to scale effectively; even minor additions of layers can deteriorate training outcomes. To address this issue, we introduce Scalable Diffusion Transformer Policy for visuomotor learning. Our proposed method, namely \methodname, introduces two modules that improve the training dynamic of Diffusion Policy and allow the network to better handle multimodal action distribution. First, we identify that \DP~suffers from large gradient issues, making the optimization of Diffusion Policy unstable. To resolve this issue, we factorize the feature embedding of observation into multiple affine layers, and integrate it into the transformer blocks. Additionally, our utilize non-causal attention which allows the policy network to \enquote{see} future actions during prediction, helping to reduce compounding errors. We demonstrate that our proposed method successfully scales the Diffusion Policy from 10 million to 1 billion parameters. This new model, named \methodname, can effectively scale up the model size with improved performance and generalization. We benchmark \methodname~across 50 different tasks from MetaWorld and find that our largest \methodname~outperforms \DP~with an average improvement of 21.6%. Across 7 real-world robot tasks, our ScaleDP demonstrates an average improvement of 36.25% over DP-T on four single-arm tasks and 75% on three bimanual tasks. We believe our work paves the way for scaling up models for visuomotor learning.

Multimodal Large Language Models (MLLMs) have showcased impressive skills in tasks related to visual understanding and reasoning. Yet, their widespread application faces obstacles due to the high computational demands during both the training and inference phases, restricting their use to a limited audience within the research and user communities. In this paper, we investigate the design aspects of Multimodal Small Language Models (MSLMs) and propose an efficient multimodal assistant named Mipha, which is designed to create synergy among various aspects: visual representation, language models, and optimization strategies. We show that without increasing the volume of training data, our Mipha-3B outperforms the state-of-the-art large MLLMs, especially LLaVA-1.5-13B, on multiple benchmarks. Through detailed discussion, we provide insights and guidelines for developing strong MSLMs that rival the capabilities of MLLMs.

Humans can communicate and observe media with different modalities, such as texts, sounds, and images. For robots to be more generalizable embodied agents, they should be capable of following instructions and perceiving the world with adaptation to diverse modalities. Current robotic learning methodologies often focus on single-modal task specification and observation, thereby limiting their ability to process rich multi-modal information. Addressing this limitation, we present an end-to-end general-purpose multi-modal system named Any-to-Policy Embodied Agents. This system empowers robots to handle tasks using various modalities, whether in combinations like text-image, audio-image, text-point cloud, or in isolation. Our innovative approach involves training a versatile modality network that adapts to various inputs and connects with policy networks for effective control. Because of the lack of existing multi-modal robotics datasets for evaluation, we assembled a comprehensive real-world dataset encompassing 30 robotic tasks. Each task in this dataset is richly annotated across multiple modalities, providing a robust foundation for assessment. We conducted extensive validation of our proposed unified modality embodied agent using several simulation benchmarks, including Franka Kitchen, Meta-World, and Maniskill2, as well as in our real-world settings. Our experiments showcase the promising capability of building embodied agents that can adapt to diverse multi-modal in a unified framework.

Transformer-based Diffusion Probabilistic Models (DPMs) have shown more potential than CNN-based DPMs, yet their extensive computational requirements hinder widespread practical applications. To reduce the computation budget of transformer-based DPMs, this work proposes the Efficient Diffusion Transformer (EDT) framework. This framework includes a lightweight-design diffusion model architecture, and a training-free Attention Modulation Matrix and its alternation arrangement in EDT inspired by human-like sketching. Additionally, we propose a token relation-enhanced masking training strategy tailored explicitly for EDT to augment its token relation learning capability. Our extensive experiments demonstrate the efficacy of EDT. The EDT framework reduces training and inference costs and surpasses existing transformer-based diffusion models in image synthesis performance, thereby achieving a significant overall enhancement. With lower FID, EDT-S, EDT-B, and EDT-XL attained speed-ups of 3.93x, 2.84x, and 1.92x respectively in the training phase, and 2.29x, 2.29x, and 2.22x respectively in inference, compared to the corresponding sizes of MDTv2.

Graph Neural Networks (GNNs) have become essential in interpreting relational data across various domains, yet, they often struggle to generalize to unseen graph data that differs markedly from training instances. In this paper, we introduce a novel framework called General Retrieval-Augmented Graph Learning (RAGraph), which brings external graph data into the general graph foundation model to improve model generalization on unseen scenarios. On the top of our framework is a toy graph vector library that we established, which captures key attributes, such as features and task-specific label information. During inference, the RAGraph adeptly retrieves similar toy graphs based on key similarities in downstream tasks, integrating the retrieved data to enrich the learning context via the message-passing prompting mechanism. Our extensive experimental evaluations demonstrate that RAGraph significantly outperforms state-of-the-art graph learning methods in multiple tasks such as node classification, link prediction, and graph classification across both dynamic and static datasets. Furthermore, extensive testing confirms that RAGraph consistently maintains high performance without the need for task-specific fine-tuning, highlighting its adaptability, robustness, and broad applicability.

Embodied agents operating in complex and uncertain environments face considerable challenges. While some advanced agents handle complex manipulation tasks with proficiency, their success often hinges on extensive training data to develop their capabilities. In contrast, humans typically rely on recalling past experiences and analogous situations to solve new problems. Aiming to emulate this human approach in robotics, we introduce the Retrieval-Augmented Embodied Agent (RAEA). This innovative system equips robots with a form of shared memory, significantly enhancing their performance. Our approach integrates a policy retriever, allowing robots to access relevant strategies from an external policy memory bank based on multi-modal inputs. Additionally, a policy generator is employed to assimilate these strategies into the learning process, enabling robots to formulate effective responses to tasks. Extensive testing of RAEA in both simulated and real-world scenarios demonstrates its superior performance over traditional methods, representing a major leap forward in robotic technology.

Humans interpret scenes by recognizing both the identities and positions of objects in their observations. For a robot to perform tasks such as \enquote{pick and place}, understanding both what the objects are and where they are located is crucial. While the former has been extensively discussed in the literature that uses the large language model to enrich the text descriptions, the latter remains underexplored. In this work, we introduce the \textit{Object-Centric Instruction Augmentation (OCI)} framework to augment highly semantic and information-dense language instruction with position cues. We utilize a Multi-modal Large Language Model (MLLM) to weave knowledge of object locations into natural language instruction, thus aiding the policy network in mastering actions for versatile manipulation. Additionally, we present a feature reuse mechanism to integrate the vision-language features from off-the-shelf pre-trained MLLM into policy networks. Through a series of simulated and real-world robotic tasks, we demonstrate that robotic manipulator imitation policies trained with our enhanced instructions outperform those relying solely on traditional language instructions.

The language-conditioned robotic manipulation aims to transfer natural language instructions into executable actions, from simple pick-and-place to tasks requiring intent recognition and visual reasoning. Inspired by the dual process theory in cognitive science, which suggests two parallel systems of fast and slow thinking in human decision-making, we introduce Robotics with Fast and Slow Thinking (RFST), a framework that mimics human cognitive architecture to classify tasks and makes decisions on two systems based on instruction types. Our RFST consists of two key components: 1) an instruction discriminator to determine which system should be activated based on the current user instruction, and 2) a slow-thinking system that is comprised of a fine-tuned vision language model aligned with the policy networks, which allows the robot to recognize user intention or perform reasoning tasks. To assess our methodology, we built a dataset featuring real-world trajectories, capturing actions ranging from spontaneous impulses to tasks requiring deliberate contemplation. Our results, both in simulation and real-world scenarios, confirm that our approach adeptly manages intricate tasks that demand intent recognition and reasoning.

Knowledge distillation (KD) can effectively compress neural networks by training a smaller network (student) to simulate the behavior of a larger one (teacher). A counter-intuitive observation is that a more expansive teacher does not make a better student, but the reasons for this phenomenon remain unclear. In this paper, we demonstrate that this is directly attributed to the presence of \textit{undistillable classes}: when trained with distillation, the teacher's knowledge of some classes is incomprehensible to the student model. We observe that while KD improves the overall accuracy, it is at the cost of the model becoming inaccurate in these undistillable classes. After establishing their widespread existence in state-of-the-art distillation methods, we illustrate their correlation with the capacity gap between teacher and student models. Finally, we present a simple Teach Less Learn More (TLLM) framework to identify and discard the undistillable classes during training. We validate the effectiveness of our approach on multiple datasets with varying network architectures. In all settings, our proposed method is able to exceed the performance of competitive state-of-the-art techniques.

In this paper, we propose a novel representation for grasping using contacts between multi-finger robotic hands and objects to be manipulated. This representation significantly reduces the prediction dimensions and accelerates the learning process. We present an effective end-to-end network, CMG-Net, for grasping unknown objects in a cluttered environment by efficiently predicting multi-finger grasp poses and hand configurations from a single-shot point cloud. Moreover, we create a synthetic grasp dataset that consists of five thousand cluttered scenes, 80 object categories, and 20 million annotations. We perform a comprehensive empirical study and demonstrate the effectiveness of our grasping representation and CMG-Net. Our work significantly outperforms the state-of-the-art for three-finger robotic hands. We also demonstrate that the model trained using synthetic data perform very well for real robots.

Accurate predicting the trajectories of moving pedestrians is a key technology in automatic driving system, which is challenging due to the complex interactions in pedestrians. Recent studies have shown that Spatio-Temporal (ST) graph has great ability to capture interactions between pedestrians. However, these methods neglect pedestrian’s limited vision and contains many invalid interactions. In order to tackle this issue, we proposed a Spatial Attention based Graph Convolutional Neural Network (SA-GCNN), which uses SA module to construct ST graph and focus on the most useful interactions. Meanwhile, SA-GCNN introduces temporal convolution module to capture temporal dependency between ST graphs. Moreover, the Graph Convolutional Network (GCN) and Temporal Convolution Network (TCN) are combined to extract graph features and decode multi-modal trajectories. Our model is trained on the widely-accepted benchmark datasets ETH and UCY. The empirical findings demonstrate that our SA-GCNN surpasses the performance of existing state-of-the-art methods used for comparison, suggests that our proposed model exhibits enhanced proficiency in capturing pedestrian interactions.

Trajectory prediction is crucial in enabling autonomous driving systems to make informed decisions, plan appropriate paths, and enhance traffic safety and efficiency. It remains an immensely challenging task due to complex interaction between vehicles, the difficulty of extracting traffic rules information, and the dynamic changes in traffic flow. Existing methods model the interactions among vehicles or extract traffic flow density features, but overlook the effects of neighboring vehicles’ movements and interactions, which contain traffic rules and the influence of surrounding traffic conditions. To achieve this, we propose a new method taking into account neighboring vehicles’ dynamic behaviors and history, allowing for a more comprehensive understanding of the traffic environment. Firstly, a novel dual-stream nearest vehicle attention mechanism method is proposed that leverages the historical state and position of the neighbors’ vehicle and captures its features. Secondly, in order to effectively encode these features, we recombine them by the multi-head attention mechanism. Lastly, in order to fusion these features and other inputs, we extract and combine the relationships between them by a self-attention mechanism. Our approach not only outperforms other methods in evaluation metrics but also demonstrates excellent results in real-world scenarios.

Spatial crowdsourcing (SC) has proven as a promising paradigm to employ human workers to collect data from diverse Point-of-Interests (PoIs) in a given area. Different from using human participants, we propose a novel air-ground SC scenario to fully take advantage of benefits brought by unmanned vehicles (UVs), including unmanned aerial vehicles (UAVs) with controllable high mobility and unmanned ground vehicles (UGVs) with abundant sensing resources. The objective is to maximize the amount of collected data, geographical fairness among all PoIs, and minimize the data loss and energy consumption, integrated as one single metric called 'efficiency'. We explicitly explore both individuality and cooperation natures of UAVs and UGVs by proposing a multi-agent deep reinforcement learning (MADRL) framework called 'h/i-MADRL'. Compatible with all multi-agent actor-critic methods, h/i-MADRL adds two novel plug-in modules: (a) h-CoPO, which models the cooperation preference among heterogeneous UAVs and UGVs; and (b) i-EOI, which extracts the UV’s individuality and encourages a better spatial division of work by adding intrinsic reward. Extensive experimental results on two real-world datasets on Purdue and NCSU campuses confirm that h/i-MADRL achieves a better exploration of both individuality and cooperation simultaneously, resulting in a better performance in terms of efficiency compared with five baselines.

The security concerns surrounding Large Language Models (LLMs) have been extensively explored, yet the safety of Multimodal Large Language Models (MLLMs) remains understudied. In this paper, we observe that Multimodal Large Language Models (MLLMs) can be easily compromised by query-relevant images, as if the text query itself were malicious. To address this, we introduce MM-SafetyBench, a comprehensive framework designed for conducting safety-critical evaluations of MLLMs against such image-based manipulations. We have compiled a dataset comprising 13 scenarios, resulting in a total of 5,040 text-image pairs. Our analysis across 12 state-of-the-art models reveals that MLLMs are susceptible to breaches instigated by our approach, even when the equipped LLMs have been safety-aligned. In response, we propose a straightforward yet effective prompting strategy to enhance the resilience of MLLMs against these types of attacks. Our work underscores the need for a concerted effort to strengthen and enhance the safety measures of open-source MLLMs against potential malicious exploits.

Attracted by the impressive power of Multimodal Large Language Models (MLLMs), the public is increasingly utilizing them to improve the efficiency of daily work. Nonetheless, the vulnerabilities of MLLMs to unsafe instructions bring huge safety risks when these models are deployed in real-world scenarios. In this paper, we systematically survey current efforts on the evaluation, attack, and defense of MLLMs' safety on images and text. We begin with introducing the overview of MLLMs on images and text and understanding of safety, which helps researchers know the detailed scope of our survey. Then, we review the evaluation datasets and metrics for measuring the safety of MLLMs. Next, we comprehensively present attack and defense techniques related to MLLMs' safety. Finally, we analyze several unsolved issues and discuss promising research directions.

This paper investigates the Neural Tangent Kernel (NTK) to search vision transformers without training. In contrast with the previous observation that NTK-based metrics can effectively predict CNNs performance at initialization, we empirically show their inefficacy in the ViT search space. We hypothesize that the fundamental feature learning preference within ViT contributes to the ineffectiveness of applying NTK to NAS for ViT. We both theoretically and empirically validate that NTK essentially estimates the ability of neural networks that learn low-frequency signals, completely ignoring the impact of high-frequency signals in feature learning. To address this limitation, we propose a new method called ViNTK that generalizes the standard NTK to the high-frequency domain by integrating the Fourier features from inputs. Experiments with multiple ViT search spaces on image classification and semantic segmentation tasks show that our method can significantly speed up search costs over prior state-of-the-art NAS for ViT while maintaining similar performance on searched architectures.

Despite the prominent success of general object detection, the performance and efficiency of Small Object Detection (SOD) are still unsatisfactory. Unlike existing works that struggle to balance the trade-off between inference speed and SOD performance, in this paper, we propose a novel Scale-aware Knowledge Distillation (ScaleKD), which transfers knowledge of a complex teacher model to a compact student model. We design two novel modules to boost the quality of knowledge transfer in distillation for SOD: 1) a scale-decoupled feature distillation module that disentangled teacher's feature representation into multi-scale embedding that enables explicit feature mimicking of the student model on small objects. 2) a cross-scale assistant to refine the noisy and uninformative bounding boxes prediction student models, which can mislead the student model and impair the efficacy of knowledge distillation. A multi-scale cross-attention layer is established to capture the multi-scale semantic information to improve the student model. We conduct experiments on COCO and VisDrone datasets with diverse types of models, i.e., two-stage and one-stage detectors, to evaluate our proposed method. Our ScaleKD achieves superior performance on general detection performance and obtains spectacular improvement regarding the SOD performance.

Detecting 3D objects from point clouds is a practical yet challenging task that has attracted increasing attention recently. In this paper, we propose a Label-Guided auxiliary training method for 3D object detection (LG3D), which serves as an auxiliary network to enhance the feature learning of existing 3D object detectors. Specifically, we propose two novel modules: a Label-Annotation-Inducer that maps annotations and point clouds in bounding boxes to task-specific representations and a Label-Knowledge-Mapper that assists the original features to obtain detection-critical representations. The proposed auxiliary network is discarded in inference and thus has no extra computational cost at test time. We conduct extensive experiments on both indoor and outdoor datasets to verify the effectiveness of our approach. For example, our proposed LG3D improves VoteNet by 2.5% and 3.1% mAP on the SUN RGB-D and ScanNetV2 datasets, respectively.

Knowledge distillation (KD) transfers the dark knowledge from cumbersome networks (teacher) to lightweight (student) networks and expects the student to achieve more promising performance than training without the teacher's knowledge. However, a counter-intuitive argument is that better teachers do not make better students due to the capacity mismatch. To this end, we present a novel adaptive knowledge distillation method to complement traditional approaches. The proposed method, named as Student Customized Knowledge Distillation (SCKD), examines the capacity mismatch between teacher and student from the perspective of gradient similarity. We formulate the knowledge distillation as a multi-task learning problem so that the teacher transfers knowledge to the student only if the student can benefit from learning such knowledge. We validate our methods on multiple datasets with various teacher-student configurations on image classification, object detection, and semantic segmentation.

Vision-based autonomous urban driving in dense traffic is quite challenging due to the complicated urban environment and the dynamics of the driving behaviors. Widely-applied methods either heavily rely on hand-crafted rules or learn from limited human experience, which makes them hard to generalize to rare but critical scenarios. In this paper, we present a novel CAscade Deep REinforcement learning framework, CADRE, to achieve model-free vision-based autonomous urban driving. In CADRE, to derive representative latent features from raw observations, we first offline train a Co-attention Perception Module (CoPM) that leverages the co-attention mechanism to learn the inter-relationships between the visual and control information from a pre-collected driving dataset. Cascaded by the frozen CoPM, we then present an efficient distributed proximal policy optimization framework to online learn the driving policy under the guidance of particularly designed reward functions. We perform a comprehensive empirical study with the CARLA NoCrash benchmark as well as specific obstacle avoidance scenarios in autonomous urban driving tasks. The experimental results well justify the effectiveness of CADRE and its superiority over the state-of-the-art by a wide margin.

The raw depth image captured by the indoor depth sensor usually has an extensive range of missing depth values due to inherent limitations such as the inability to perceive transparent objects and limited distance range. The incomplete depth map burdens many downstream vision tasks, and a rising number of depth completion methods have been proposed to alleviate this issue. While most existing methods can generate accurate dense depth maps from sparse and uniformly sampled depth maps, they are not suitable for complementing the large contiguous regions of missing depth values, which is common and critical. In this paper, we design a novel two-branch end-to-end fusion network, which takes a pair of RGB and incompleted depth images as input to predict a dense and completed depth map. The first branch employs an encoder-decoder structure to regress the local dense depth values from the raw depth map, with the help of local guidance information extracted from the RGB image. In the other branch, we propose an RGB-depth fusion GAN to transfer the RGB image to the fine-grained textured depth map. We adopt adaptive fusion modules named W-AdaIN to propagate the features across the two branches, and we append a confidence fusion head to fuse the two outputs of the branches for the final depth map. Extensive experiments on NYU-Depth V2 and SUN RGB-D demonstrate that our proposed method clearly improves the depth completion performance, especially in a more realistic setting of indoor environments with the help of the pseudo depth map.

Semantic context has raised concerns in semantic segmentation. In most cases, it is applied to guide feature learning. Instead, this paper applies it to extract the semantic representation, which records the global feature information of each category with a memory tensor. Specifically, we propose a novel semantic representation (SR) module, which consists of semantic embedding (SE) and semantic attention (SA) blocks. The SE block adaptively embeds features into the semantic representation by calculating the memory similarity, and the SA block aggregates the embedded features with semantic attention. The main advantages of the SR module lie in three aspects: i) it enhances the representation ability of semantic context by employing global (cross-image) semantic information; ii) it improves the consistency of intraclass features by aggregating global features of the same categories; and iii) it can be extended to build a semantic representation refinement network (SRRNet) by iteratively applying the SR module across multiple scales, shrinking the semantic gap and enhancing the structural reasoning of the model. Extensive experiments demonstrate that our method significantly improves the segmentation results and achieves superior performance on the PASCAL VOC 2012, Cityscapes, and PASCAL Context datasets.

Channel pruning has been widely studied as a prevailing method that effectively reduces both computational cost and memory footprint of the original network while keeping a comparable accuracy performance. Though great success has been achieved in channel pruning for 2D image-based convolutional networks (CNNs), existing works seldom extend the channel pruning methods to 3D point-based neural networks (PNNs). Directly implementing the 2D CNN channel pruning methods to PNNs undermine the performance of PNNs because of the different representations of 2D images and 3D point clouds as well as the network architecture disparity. In this paper, we proposed CP^3, which is a Channel Pruning Plug-in for Point-based network. CP^3 is elaborately designed to leverage the characteristics of point clouds and PNNs in order to enable 2D channel pruning methods for PNNs. Specifically, it presents a coordinate-enhanced channel importance metric to reflect the correlation between dimensional information and individual channel features, and it recycles the discarded points in PNN's sampling process and reconsiders their potentially-exclusive information to enhance the robustness of channel pruning. Experiments on various PNN architectures show that CP^3 constantly improves state-of-the-art 2D CNN pruning approaches on different point cloud tasks. For instance, our compressed PointNeXt-S on ScanObjectNN achieves an accuracy of 88.52% with a pruning rate of 57.8%, outperforming the baseline pruning methods with an accuracy gain of 1.94%.

Estimating 6D poses and reconstructing 3D shapes of objects in open-world scenes from RGB-depth image pairs is challenging. Many existing methods rely on learning geometric features that correspond to specific templates while disregarding shape variations and pose differences among objects in the same category. As a result, these methods underperform when handling unseen object instances in complex environments. In contrast, other approaches aim to achieve category-level estimation and reconstruction by leveraging normalized geometric structure priors, but the static prior-based reconstruction struggles with substantial intra-class variations. To solve these problems, we propose the DTF-Net, a novel framework for pose estimation and shape reconstruction based on implicit neural fields of object categories. In DTF-Net, we design a deformable template field to represent the general category-wise shape latent features and intra-category geometric deformation features. The field establishes continuous shape correspondences, deforming the category template into arbitrary observed instances to accomplish shape reconstruction. We introduce a pose regression module that shares the deformation features and template codes from the fields to estimate the accurate 6D pose of each object in the scene. We integrate a multi-modal representation extraction module to extract object features and semantic masks, enabling end-to-end inference. Moreover, during training, we implement a shape-invariant training strategy and a viewpoint sampling method to further enhance the model's capability to extract object pose features. Extensive experiments on the REAL275 and CAMERA25 datasets demonstrate the superiority of DTF-Net in both synthetic and real scenes. Furthermore, we show that DTF-Net effectively supports grasping tasks with a real robot arm.

Reconstructing real-world objects and estimating their movable joint structures are pivotal technologies within the field of robotics. Previous research has predominantly focused on supervised approaches, relying on extensively annotated datasets to model articulated objects within limited categories. However, this approach falls short of effectively addressing the diversity present in the real world. To tackle this issue, we propose a self-supervised interaction perception method, referred to as SM, which leverages multi-view RGB images captured before and after interaction to model articulated objects, identify the movable parts, and infer the parameters of their rotating joints. By constructing 3D geometries and textures from the captured 2D images, SM achieves integrated optimization of movable part and joint parameters during the reconstruction process, obviating the need for annotations. Furthermore, we introduce the MMArt dataset, an extension of PartNet-Mobility, encompassing multi-view and multi-modal data of articulated objects spanning diverse categories. Evaluations demonstrate that SM surpasses existing benchmarks across various categories and objects, while its adaptability in real-world scenarios has been thoroughly validated.

Generative Adversarial Imitation Learning (GAIL) stands as a cornerstone approach in imitation learning. This paper investigates the gradient explosion in two types of GAIL: GAIL with deterministic policy (DE-GAIL) and GAIL with stochastic policy (ST-GAIL). We begin with the observation that the training can be highly unstable for DE-GAIL at the beginning of the training phase and end up divergence. Conversely, the ST-GAIL training trajectory remains consistent, reliably converging. To shed light on these disparities, we provide an explanation from a theoretical perspective. By establishing a probabilistic lower bound for GAIL, we demonstrate that gradient explosion is an inevitable outcome for DE-GAIL due to occasionally large expert-imitator policy disparity, whereas ST-GAIL does not have the issue with it. To substantiate our assertion, we illustrate how modifications in the reward function can mitigate the gradient explosion challenge. Finally, we propose CREDO, a simple yet effective strategy that clips the reward function during the training phase, allowing the GAIL to enjoy high data efficiency and stable trainability.

Neural network compression techniques, such as knowledge distillation (KD) and network pruning, have received increasing attention. Recent work `Prune, then Distill' reveals that a pruned student-friendly teacher network can benefit the performance of KD. However, the conventional teacher-student pipeline, which entails cumbersome pre-training of the teacher and complicated compression steps, makes pruning with KD less efficient. In addition to compressing models, recent compression techniques also emphasize the aspect of efficiency. Early pruning demands significantly less computational cost in comparison to the conventional pruning methods as it does not require a large pre-trained model. Likewise, a special case of KD, known as self-distillation (SD), is more efficient since it requires no pre-training or student-teacher pair selection. This inspires us to collaborate early pruning with SD for efficient model compression. In this work, we propose the framework named Early Pruning with Self-Distillation (EPSD), which identifies and preserves distillable weights in early pruning for a given SD task. EPSD efficiently combines early pruning and self-distillation in a two-step process, maintaining the pruned network's trainability for compression. Instead of a simple combination of pruning and SD, EPSD enables the pruned network to favor SD by keeping more distillable weights before training to ensure better distillation of the pruned network. We demonstrated that EPSD improves the training of pruned networks, supported by visual and quantitative analyses. Our evaluation covered diverse benchmarks (CIFAR-10/100, Tiny-ImageNet, full ImageNet, CUB-200-2011, and Pascal VOC), with EPSD outperforming advanced pruning and SD techniques.

The vision transformer is a model that breaks down each image into a sequence of tokens with a fixed length and processes them similarly to words in natural language processing. Although increasing the number of tokens typically results in better performance, it also leads to a considerable increase in computational cost. Motivated by the saying 'A picture is worth a thousand words,'we propose an innovative approach to accelerate the ViT model by shortening long images. Specifically, we introduce a method for adaptively assigning token length for each image at test time to accelerate inference speed. First, we train a Resizable-ViT (ReViT) model capable of processing input with diverse token lengths. Next, we extract token-length labels from ReViT that indicate the minimum number of tokens required to achieve accurate predictions. We then use these labels to train a lightweight Token-Length Assigner (TLA) that allocates the optimal token length for each image during inference. The TLA enables ReViT to process images with the minimum sufficient number of tokens, reducing token numbers in the ViT model and improving inference speed. Our approach is general and compatible with modern vision transformer architectures, significantly reducing computational costs. We verified the effectiveness of our methods on multiple representative ViT models on image classification and action recognition.

Generative adversarial imitation learning (GAIL) learns an optimal policy by expert demonstrations from the environment with unknown reward functions. Different from existing works that studied the generalization of reward function classes or discriminator classes, we focus on policy classes. This paper investigates the generalization and computation for policy classes of GAIL. Specifically, our contributions lie in: 1) We prove that the generalization is guaranteed in GAIL when the complexity of policy classes is properly controlled. 2) We provide an off-policy framework called the two-stage stochastic gradient (TSSG), which can efficiently solve GAIL based on the soft policy iteration and attain the sublinear convergence rate to a stationary solution. The comprehensive numerical simulations are illustrated in MuJoCo environments.

Generative adversarial imitation learning (GAIL) learns an optimal policy by expert demonstrations from the environment with unknown reward functions. Different from existing works that studied the generalization of reward function classes or discriminator classes, we focus on policy classes. This paper investigates the generalization and computation for policy classes of GAIL. Specifically, our contributions lie in: 1) We prove that the generalization is guaranteed in GAIL when the complexity of policy classes is properly controlled. 2) We provide an off-policy framework called the two-stage stochastic gradient (TSSG), which can efficiently solve GAIL based on the soft policy iteration and attain the sublinear convergence rate to a stationary solution. The comprehensive numerical simulations are illustrated in MuJoCo environments.