Bimanual dexterous manipulation is a critical yet underexplored area in robotics. Its high-dimensional action space and inherent task complexity present significant challenges for policy learning, and the limited task diversity in existing benchmarks hinders general-purpose skill development. Existing approaches largely depend on reinforcement learning, often constrained by intricately designed reward functions tailored to a narrow set of tasks. In this work, we present a novel approach for efficiently learning diverse bimanual dexterous skills from abundant human demonstrations. Specifically, we introduce BiDexHD, a framework that unifies task construction from existing bimanual datasets and employs teacher-student policy learning to address all tasks. The teacher learns state-based policies using a general two-stage reward function across tasks with shared behaviors, while the student distills the learned multi-task policies into a vision-based policy. With BiDexHD, scalable learning of numerous bimanual dexterous skills from auto-constructed tasks becomes feasible, offering promising advances toward universal bimanual dexterous manipulation. Experiments on TACO tool-using dataset spanning 141 tasks across 6 categories demonstrate a task fulfillment rate of 74.59% on trained tasks and 51.07% on unseen tasks. We further transfer BiDexHD to 11 ARCTIC collaborative tasks and achieve an average of 80.49% task fulfillment rate on trained tasks and 65.99% on unseen task. All empirical results demonstrate the effectiveness and competitive zero-shot generalization capabilities of BiDexHD.

Large Language Models (LLMs) are increasingly deployed in real-world applications that demand complex reasoning.To track progress, robust benchmarks are required to evaluate their capabilities beyond superficial pattern recognition.However, current LLM reasoning benchmarks often face challenges such as insufficient interpretability, performance saturation or data contamination.To address these challenges, we introduce GAMEBOT (GAME Battle of Tactics), a gaming arena designed for rigorous and transparent assessment of LLM reasoning capabilities.GAMEBOT decomposes complex reasoning in games into predefined modular subproblems.This decomposition allows us to design a suite of Chain-of-Thought (CoT) prompts that leverage domain knowledge to guide LLMs in addressing these subproblems before action selection.Furthermore, we develop a suite of rule-based algorithms to generate ground truth for these subproblems, enabling rigorous validation of the LLMs' intermediate reasoning steps.This approach facilitates evaluation of both the quality of final actions and the accuracy of the underlying reasoning process.GAMEBOT also naturally alleviates the risk of data contamination through dynamic games and head-to-head LLM competitions.We benchmark 17 prominent LLMs across eight games, encompassing various strategic abilities and game characteristics.Our results suggest that GAMEBOT presents a significant challenge, even when LLMs are provided with detailed CoT prompts.Project page: https://visual-ai.github.io/gamebot

Multimodal large language models (MLLMs), built on large-scale pre-trained vision towers and language models, have shown great capabilities in multimodal understanding. However, most existing MLLMs are trained on single-turn vision question-answering tasks, which do not accurately reflect real-world human conversations. In this paper, we introduce MMDiag, a new large-scale multi-turn multimodal dialogue dataset. This dataset is collaboratively generated through deliberately designed rules and GPT assistance, featuring complex dialogues with contextual dependencies that force models to track, ground, and recall information across multiple turns and disparate visual regions. MMDiag serves as a strong benchmark for multi-turn multimodal dialogue learning and brings more challenges to the grounding and reasoning capabilities of MLLMs. Further, inspired by human vision processing we present DiagNote, equipped with multimodal grounding and reasoning capabilities. DiagNote adopts a novel dual-module architecture that explicitly separates reasoning from grounding: a reasoning module (Deliberate) performs step-by-step Chain-of-Thought, while a grounding module (Gaze) provides precise visual focus by predicting bounding box annotations. These modules interact iteratively, enabling DiagNote to dynamically refine its understanding. We empirically demonstrate the advantages of DiagNote in both grounding and jointly processing and reasoning with vision and language information over existing MLLMs.

We present VideoOrion, a Video Large Language Model (Video-LLM) that explicitly captures the key semantic information in videos - the spatial-temporal dynamics of objects throughout the videos. VideoOrion employs expert vision models to extract object dynamics through a detect-segment-track pipeline, encoding them into a set of object tokens by aggregating spatial-temporal object features. Our method addresses the persistent challenge in Video-LLMs of efficiently compressing high-dimensional video data into semantic tokens that are comprehensible to LLMs. Compared to prior methods which resort to downsampling the original video or aggregating visual tokens using resamplers, leading to information loss and entangled semantics, VideoOrion not only offers a more natural and efficient way to derive compact, disentangled semantic representations but also enables explicit object modeling of video content with minimal computational cost. Moreover, the introduced object tokens naturally allow VideoOrion to accomplish video-based referring tasks. Experimental results show that VideoOrion can learn to make good use of the object tokens, and achieves competitive results on both general video question answering and video-based referring benchmarks.

Reinforcement learning with verifiable outcome rewards (RLVR) has effectively scaled up chain-of-thought (CoT) reasoning in large language models (LLMs). Yet, its efficacy in training vision-language model (VLM) agents for goal-directed action reasoning in visual environments is less established. This work investigates this problem through extensive experiments on complex card games, such as 24 points, and embodied tasks from ALFWorld. We find that when rewards are based solely on action outcomes, RL fails to incentivize CoT reasoning in VLMs, instead leading to a phenomenon we termed thought collapse, characterized by a rapid loss of diversity in the agent's thoughts, state-irrelevant and incomplete reasoning, and subsequent invalid actions, resulting in negative rewards. To counteract thought collapse, we highlight the necessity of process guidance and propose an automated corrector that evaluates and refines the agent's reasoning at each RL step. This simple and scalable GTR (Guided Thought Reinforcement) framework trains reasoning and action simultaneously without the need for dense, per-step human labeling. Our experiments demonstrate that GTR significantly enhances the performance and generalization of the LLaVA-7b model across various visual environments, achieving 3-5 times higher task success rates compared to SoTA models with notably smaller model sizes.

Multimodal large language models (MLLMs) have made significant progress in vision-language understanding, yet effectively aligning different modalities remains a fundamental challenge. We present a framework that unifies multimodal understanding by applying byte-pair encoding to visual tokens. Unlike conventional approaches that rely on modality-specific encoders, our method directly incorporates structural information into visual tokens, mirroring successful tokenization strategies in text-only language models. We introduce a priority-guided encoding scheme that considers both frequency and spatial consistency, coupled with a multi-stage training procedure based on curriculum-driven data composition. These enhancements enable the transformer model to better capture cross-modal relationships and reason with visual information. Comprehensive experiments demonstrate improved performance across diverse vision-language tasks. By bridging the gap between visual and textual representations, our approach contributes to the advancement of more capable and efficient multimodal foundation models.

Multimodal Large Language Models have made significant strides in integrating visual and textual information, yet they often struggle with effectively aligning these modalities. We introduce a novel image tokenizer that bridges this gap by applying the principle of Byte-Pair Encoding (BPE) to visual data. Unlike conventional approaches that rely on separate visual encoders, our method directly incorporates structural prior information into image tokens, mirroring the successful tokenization strategies used in text-only Large Language Models. This innovative approach enables Transformer models to more effectively learn and reason across modalities. Through theoretical analysis and extensive experiments, we demonstrate that our BPE Image Tokenizer significantly enhances MLLMs' multimodal understanding capabilities, even with limited training data. Leveraging this method, we develop Being-VL-0, a model that demonstrates superior performance across various benchmarks and shows promising scalability, potentially paving the way for more efficient and capable multimodal foundation models. For further details, visit our website https://github.com/BeingBeyond/Being-VL-0.

Universal dexterous grasping across diverse objects presents a fundamental yet formidable challenge in robot learning. Existing approaches using reinforcement learning (RL) to develop policies on extensive object datasets face critical limitations, including complex curriculum design for multi-task learning and limited generalization to unseen objects. To overcome these challenges, we introduce ResDex, a novel approach that integrates residual policy learning with a mixture-of-experts (MoE) framework. ResDex is distinguished by its use of geometry-agnostic base policies that are efficiently acquired on individual objects and capable of generalizing across a wide range of unseen objects. Our MoE framework incorporates several base policies to facilitate diverse grasping styles suitable for various objects. By learning residual actions alongside weights that combine these base policies, ResDex enables efficient multi-task RL for universal dexterous grasping. ResDex achieves state-of-the-art performance on the DexGraspNet dataset comprising 3,200 objects with an 88.8% success rate. It exhibits no generalization gap with unseen objects and demonstrates superior training efficiency, mastering all tasks within only 12 hours on a single GPU. For further details and videos, visit our project page.

Skill learning from language instructions is a critical challenge in developing intelligent agents that can generalize across diverse tasks and follow complex human instructions. Hierarchical methods address this by decomposing the learning problem into multiple levels, where the high-level and low-level policies are mediated through a latent plan space. Effective modeling and learning of this latent plan space are key to enabling robust and interpretable skill learning. In this paper, we introduce LADS, a hierarchical approach that learns language-conditioned discrete latent plans through semantic skill abstractions. Our method decouples the learning of the latent plan space from the language-conditioned high-level policy to improve training stability. First, we incorporate a trajectory encoder to learn a discrete latent space with the low-level policy, regularized by language instructions. Next, we model the high-level policy as a categorical distribution over these discrete latent plans to capture the multi-modality of the dataset. Through experiments in simulated control environments, we demonstrate that LADS outperforms state-of-the-art methods in both skill learning and compositional generalization.

The video-conditioned policy takes prompt videos of the desired tasks as a condition and is regarded for its prospective generalizability. Despite its promise, training a video-conditioned policy is non-trivial due to the need for abundant demonstrations. In some tasks, the expert rollouts are merely available as videos, and costly and time-consuming efforts are required to annotate action labels. To address this, we explore training video-conditioned policy on a mixture of demonstrations and unlabeled expert videos to reduce reliance on extensive manual annotation. We introduce the Joint Embedding Predictive Transformer (JEPT) to learn a video-conditioned policy through sequence modeling. JEPT is designed to jointly learn visual transition prediction and inverse dynamics. The visual transition is captured from both demonstrations and expert videos, on the basis of which the inverse dynamics learned from demonstrations is generalizable to the tasks without action labels. Experiments on a series of simulated visual control tasks evaluate that JEPT can effectively leverage the mixture dataset to learn a generalizable policy. JEPT outperforms baselines in the tasks without action-labeled data and unseen tasks. We also experimentally reveal the potential of JEPT as a simple visual priors injection approach to enhance the video-conditioned policy.

We explore cross-domain offline reinforcement learning (RL) where offline datasets from another domain can be accessed to facilitate policy learning. However, the underlying environments of the two datasets may have dynamics mismatches, incurring inferior performance when simply merging the data of two domains. Existing methods mitigate this issue by training domain classifiers, using contrastive learning methods, etc. Nevertheless, they still rely on a large amount of target domain data to function well. Instead, we address this problem by establishing a concrete performance bound of a policy given datasets from two domains. Motivated by the theoretical insights, we propose to align transitions in the two datasets using optimal transport and selectively share source domain samples, without training any neural networks. This enables reliable data filtering even given a few target domain data. Additionally, we introduce a dataset regularization term that ensures the learned policy remains within the scope of the target domain dataset, preventing it from being biased towards the source domain data. Consequently, we propose the Optimal Transport Data Filtering (dubbed OTDF) method and examine its effectiveness by conducting extensive experiments across various dynamics shift conditions (e.g., gravity shift), given limited target domain data. It turns out that OTDF exhibits superior performance on many tasks and dataset qualities, often surpassing prior strong baselines by a large margin.

In this paper, we study the problem of video-conditioned policy learning. While previous works mostly focus on learning policies that perform a single skill specified by the given video, we take a step further and aim to learn a policy that can perform multiple skills according to the given video, and generalize to unseen videos by recombining these skills. To solve this problem, we propose our algorithm, Watch-Less-Do-More, an information bottleneck-based imitation learning framework for implicit skill discovery and video-conditioned policy learning. In our method, an information bottleneck objective is employed to control the information contained in the video representation, ensuring that it only encodes information relevant to the current skill (Watch-Less). By discovering potential skills from training videos, the learned policy is able to recombine them and generalize to unseen videos to achieve compositional generalization (Do-More). To evaluate our method, we perform extensive experiments in various environments and show that our algorithm substantially outperforms baselines (up to 2x) in terms of compositional generalization ability.

Dexterous hands exhibit significant potential for complex real-world grasping tasks. While recent studies have primarily focused on learning policies for specific robotic hands, the development of a universal policy that controls diverse dexterous hands remains largely unexplored. In this work, we study the learning of cross-embodiment dexterous grasping policies using reinforcement learning (RL). Inspired by the capability of human hands to control various dexterous hands through teleoperation, we propose a universal action space based on the human hand's eigengrasps. The policy outputs eigengrasp actions that are then converted into specific joint actions for each robot hand through a retargeting mapping. We simplify the robot hand's proprioception to include only the positions of fingertips and the palm, offering a unified observation space across different robot hands. Our approach demonstrates an 80\% success rate in grasping objects from the YCB dataset across four distinct embodiments using a single vision-based policy. Additionally, our policy exhibits zero-shot generalization to two previously unseen embodiments and significant improvement in efficient finetuning. For further details and videos, visit our project page (https://sites.google.com/view/crossdex).

MLLM agents demonstrate potential for complex embodied tasks by retrieving multimodal task-relevant trajectory data. However, current retrieval methods primarily focus on surface-level similarities of textual or visual cues in trajectories, neglecting their effectiveness for the specific task at hand. To address this issue, we propose a novel method, MART, which enhances the performance of embodied agents by utilizing interaction data to fine-tune an MLLM retriever based on preference learning, such that the retriever fully considers the effectiveness of trajectories and prioritize them for unseen tasks. We also introduce Trajectory Abstraction, a mechanism that leverages MLLMs' summarization capabilities to represent trajectories with fewer tokens while preserving key information, enabling agents to better comprehend milestones in the trajectory. Experimental results across various environments demonstrate our method significantly improves task success rates in unseen scenes compared to baseline methods. This work presents a new paradigm for multimodal retrieval in embodied agents, by fine-tuning a general-purpose MLLM as the retriever to assess trajectory effectiveness. All the code for benchmark tasks, simulator modifications and the MLLM retriever is available at https://github.com/PKU-RL/MART.

Cooperative Multi-Agent Reinforcement Learning (MARL) has become a critical tool for addressing complex real-world problems. However, off-policy MARL methods, which rely on joint Q-functions, face significant scalability challenges due to the exponentially growing joint action space. In this work, we highlight a critical yet often overlooked issue: erroneous Q-target estimation, primarily caused by extrapolation error. Our analysis reveals that this error becomes increasingly severe as the number of agents grows, leading to unique challenges in MARL due to its expansive joint action space and the decentralized execution paradigm. To address these challenges, we propose a suite of techniques tailored for off-policy MARL, including annealed multi-step bootstrapping, averaged Q-targets, and restricted action representation. Experimental results demonstrate that these methods effectively mitigate erroneous estimations, yielding substantial performance improvements in challenging benchmarks such as SMAC, SMACv2, and Google Research Football.

Despite their success in specific scenarios, existing foundation agents still struggle to generalize across various virtual scenarios, mainly due to the dramatically different encapsulations of environments with manually designed observation and action spaces. To handle this issue, we propose the General Computer Control (GCC) setting to restrict foundation agents to interact with software through the most unified and standardized interface, i.e., using screenshots as input and keyboard and mouse actions as output. We introduce Cradle, a modular and flexible LMM-powered framework, as a preliminary attempt towards GCC. Enhanced by six key modules, Information Gathering, Self-Reflection, Task Inference, Skill Curation, Action Planning, and Memory, Cradle is able to understand input screenshots and output executable code for low-level keyboard and mouse control after high-level planning and information retrieval, so that Cradle can interact with any software and complete long-horizon complex tasks without relying on any built-in APIs. Experimental results show that Cradle exhibits remarkable generalizability and impressive performance across four previously unexplored commercial video games (Red Dead Redemption 2, Cities:Skylines, Stardew Valley and Dealer’s Life 2), five software applications (Chrome, Outlook, Feishu, Meitu and CapCut), and a comprehensive benchmark, OSWorld. With a unified interface to interact with any software, Cradle greatly extends the reach of foundation agents thus paving the way for generalist agents.

Inspired by the recent success of LLMs, the field of human motion understanding has increasingly shifted toward developing large motion models. Despite some progress, current efforts remain far from achieving truly generalist models, primarily due to the lack of massive high-quality data. To address this gap, we present MotionLib, the first million-level dataset for motion generation, which is at least 15$\times$ larger than existing counterparts and enriched with hierarchical text descriptions. Using MotionLib, we train a large motion model named Being-M0, demonstrating robust performance across a wide range of human activities, including unseen ones. Through systematic investigation, for the first time, we highlight the importance of scaling both data and model size for advancing motion generation, along with key insights to achieve this goal. To better integrate the motion modality, we propose Motionbook, an innovative motion encoding approach including (1) a compact yet lossless feature to represent motions; (2) a novel 2D lookup-free motion tokenizer that preserves fine-grained motion details while expanding codebook capacity, significantly enhancing the representational power of motion tokens. We believe this work lays the groundwork for developing more versatile and powerful motion generation models in the future. For further details, visit https://beingbeyond.github.io/Being-M0/.

The in-context learning ability of Transformer models has brought new possibilities to visual navigation. In this paper, we focus on a novel video navigation setting, where an in-context navigation policy needs to be learned purely from videos in an offline manner, without access to the actual environment. For this setting, we propose Navigate Only Look Once (NOLO), a method for learning a navigation policy that possesses the in-context ability and adapts to new scenes by taking corresponding context videos as input without finetuning or re-training. To enable learning from videos, we first propose a pseudo action labeling procedure using optical flow to recover the action label from egocentric videos. Then, offline reinforcement learning is applied to learn the navigation policy. Through extensive experiments on different scenes both in simulation and the real world, we show that our algorithm outperforms baselines by a large margin, which demonstrates the effectiveness of the learned policy in in-context learning. For videos and more information, visit our project page.

XiaoPeng Yu, Wanpeng Zhang, Zongqing Lu. Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 2025.

Recent advances in large multimodal models have significantly advanced video comprehension, yet their performance remains limited in first-person scenarios. The interactive nature of egocentric videos is critical for applications like embodied intelligence, but introduces complex visual contexts that conventional models struggle to capture. To bridge this gap, we introduce OpenMMEgo with innovations across three dimensions: data, model, and training strategy. To provide rich spatiotemporal visual knowledge, we curate a large-scale, high-quality dataset named OME10M, comprising over 8.2M egocentric video QA pairs synthesized from Ego4D series. We also establish OMEBench, a comprehensive benchmark for rigorous egocentric understanding assessment. To alleviate the frequent viewpoint shifts inherent in egocentric videos, we implement semantic-aware visual token compression. Further, a curriculum learning strategy is complemented to foster stable learning across various data complexities. OpenMMEgo consistently improves the performance of LMMs on egocentric benchmarks without sacrificing general video understanding performance. Notably, Qwen2.5-VL tuned with OpenMMEgo substantially outperforms other models of the same size in egocentric video understanding. The data, weights and training code will be put at https://github.com/BeingBeyond/OpenMMEgo.

Achieving general agile whole-body control on humanoid robots remains a major challenge due to diverse motion demands and data conflicts. While existing frameworks excel in training single motion-specific policies, they struggle to generalize across highly varied behaviors due to conflicting control requirements and mismatched data distributions. In this work, we propose BumbleBee (BB), an expert-generalist learning framework that combines motion clustering and sim-to-real adaptation to overcome these challenges. BB first leverages an autoencoder-based clustering method to group behaviorally similar motions using motion features and motion descriptions. Expert policies are then trained within each cluster and refined with real-world data through iterative delta action modeling to bridge the sim-to-real gap. Finally, these experts are distilled into a unified generalist controller that preserves agility and robustness across all motion types. Experiments on two simulations and a real humanoid robot demonstrate that BB achieves state-of-the-art general whole-body control, setting a new benchmark for agile, robust, and generalizable humanoid performance in the real world.

Planning under opponent uncertainty is a fundamental challenge in multi-agent environments, where an agent must act while inferring the hidden policies of its opponents. Existing type-based methods rely on manually defined behavior classes and struggle to scale, while model-free approaches are sample-inefficient and lack a principled way to incorporate uncertainty into planning. We propose Quantized Opponent Models (QOM), which learn a compact catalog of opponent types via a quantized autoencoder and maintain a Bayesian belief over these types online. This posterior supports both a belief-weighted meta-policy and a Monte-Carlo planning algorithm that directly integrates uncertainty, enabling real-time belief updates and focused exploration. Experiments show that QOM achieves superior performance with lower search cost, offering a tractable and effective solution for belief-aware planning.

Egocentric hand-object motion generation is crucial for immersive AR/VR and robotic imitation but remains challenging due to unstable viewpoints, self-occlusions, perspective distortion, and noisy ego-motion. Existing methods rely on predefined 3D object priors, limiting generalization to novel objects, which restricts their generalizability to novel objects. Meanwhile, recent multimodal approaches suffer from ambiguous generation from abstract textual cues, intricate pipelines for modeling 3D hand-object correlation, and compounding errors in open-loop prediction. We propose **MEgoHand**, a multimodal framework that synthesizes physically plausible hand-object interactions from egocentric RGB, text, and initial hand pose. MEgoHand introduces a bi-level architecture: a high-level “cerebrum” leverages a vision language model (VLM) to infer motion priors from visual-textual context and a monocular depth estimator for object-agnostic spatial reasoning, while a low-level DiT-based flow-matching policy generates fine-grained trajectories with temporal orthogonal filtering to enhance stability. To address dataset inconsistency, we design a dataset curation paradigm with an Inverse MANO Retargeting Network and Virtual RGB-D Renderer, curating a unified dataset of **3.35M** RGB-D frames, **24K** interactions, and **1.2K** objects. Extensive experiments across **five** in-domain and **two** cross-domain datasets demonstrate the effectiveness of MEgoHand, achieving substantial reductions in wrist translation error (**86.9%**) and joint rotation error (**34.1%**), highlighting its capacity to accurately model fine-grained hand joint structures and generalize robustly across diverse scenarios.

We study building embodied agents for open-ended creative tasks. While existing methods build instruction-following agents that can perform diverse open-ended tasks, none of them demonstrates creativity – the ability to give novel and diverse solutions implicit in the language instructions. This limitation comes from their inability to convert abstract language instructions into concrete goals and perform long-horizon planning for such complicated goals. Given the observation that humans perform creative tasks with imagination, we propose a class of solutions, where the controller is enhanced with an imaginator generating detailed imaginations of task outcomes conditioned on language instructions. We introduce several approaches to implementing the components of creative agents. We implement the imaginator with either a large language model for textual imagination or a diffusion model for visual imagination. The controller can either be a behavior-cloning policy or a pre-trained foundation model generating executable codes in the environment. We benchmark creative tasks with the challenging open-world game Minecraft, where the agents create diverse buildings given free-form language instructions. We propose novel evaluation metrics for open-ended creative tasks utilizing GPT-4V, which holds many advantages over existing metrics. We perform a detailed experimental analysis of creative agents, showing that creative agents are the first AI agents accomplishing diverse building creation in the survival mode of Minecraft. Our benchmark and models are open-source for future research on creative agents (https://github.com/PKU-RL/Creative-Agents).

Fully decentralized learning, where the global information, *i.e.*, the actions of other agents, is inaccessible, is a fundamental challenge in cooperative multi-agent reinforcement learning. However, the convergence and optimality of most decentralized algorithms are not theoretically guaranteed, since the transition probabilities are non-stationary as all agents are updating policies simultaneously. To tackle this challenge, we propose \textit{best possible operator}, a novel decentralized operator, and prove that the policies of cooperative agents will converge to the optimal joint policy if each agent independently updates its individual state-action value by the operator when there is only one optimal joint policy. Further, to make the update more efficient and practical, we simplify the operator and prove that the convergence and optimality still hold with the simplified one. By instantiating the simplified operator, the derived fully decentralized algorithm, *best possible Q-learning* (BQL), does not suffer from non-stationarity. Empirically, we show that BQL achieves remarkable improvement over baselines in a variety of cooperative multi-agent tasks.

The study of emergent communication has been dedicated to interactive artificial intelligence. While existing work focuses on communication about single objects or complex image scenes, we argue that communicating relationships between multiple objects is important in more realistic tasks, but understudied. In this paper, we try to fill this gap and focus on emergent communication about positional relationships between two objects. We train agents in the referential game where observations contain two objects, and find that generalization is the major problem when the positional relationship is involved. The key factor affecting the generalization ability of the emergent language is the input variation between Speaker and Listener, which is realized by a random image generator in our work. Further, we find that the learned language can generalize well in a new multi-step MDP task where the positional relationship describes the goal, and performs better than raw-pixel images as well as pre-trained image features, verifying the strong generalization ability of discrete sequences. We also show that language transfer from the referential game performs better in the new task than learning language directly in this task, implying the potential benefits of pre-training in referential games. All in all, our experiments demonstrate the viability and merit of having agents learn to communicate positional relationships between multiple objects through emergent communication.

Exploration in decentralized cooperative multi-agent reinforcement learning faces two challenges. One is that the novelty of global states is unavailable, while the novelty of local observations is biased. The other is how agents can explore in a coordinated way. To address these challenges, we propose MACE, a simple yet effective multi-agent coordinated exploration method. By communicating only local novelty, agents can take into account other agents' local novelty to approximate the global novelty. Further, we newly introduce weighted mutual information to measure the influence of one agent's action on other agents' accumulated novelty. We convert it as an intrinsic reward in hindsight to encourage agents to exert more influence on other agents' exploration and boost coordinated exploration. Empirically, we show that MACE achieves superior performance in three multi-agent environments with sparse rewards.

Reinforcement learning with natural language action space often suffers from the curse of dimensionality due to the combinatorial nature of the natural language.Previous research leverages pretrained language models to capture action semantics and reduce the size of the action space.However, since pretrained models are typically trained on general corpora, there can be an unpredictable mismatch between the priors encoded in pretrained models and the characteristics of the specific RL environment.To address this issue, we propose Mutual-Information Regularized Policy Optimization, MIPO.MIPO enables implicit and dynamic reduction of the action space.Starting from the prior provided by the pretrained language model, our method dynamically adjusts the prior during the learning process based on the guidance of mutual information regularization.Theoretically, we demonstrate that this policy optimization process leads to the monotonic improvement on the mutual-information regularized RL objective.Empirically, we conduct experiments in various environments and demonstrate the effectiveness of MIPO.

Offline reinforcement learning (RL) has attracted much attention due to its ability in learning from static offline datasets and eliminating the need of interacting with the environment. Nevertheless, the success of offline RL relies heavily on the offline transitions annotated with reward labels. In practice, we often need to hand-craft the reward function, which is sometimes difficult, labor-intensive, or inefficient. To tackle this challenge, we set our focus on the offline imitation learning (IL) setting, and aim at getting a reward function based on the expert data and unlabeled data. To that end, we propose a simple yet effective search-based offline IL method, tagged SEABO. SEABO allocates a larger reward to the transition that is close to its closest neighbor in the expert demonstration, and a smaller reward otherwise, all in an unsupervised learning manner. Experimental results on a variety of D4RL datasets indicate that SEABO can achieve competitive performance to offline RL algorithms with ground-truth rewards, given only a single expert trajectory, and can outperform prior reward learning and offline IL methods across many tasks. Moreover, we demonstrate that SEABO also works well if the expert demonstrations contain only observations. Our code is publicly available at https://github.com/dmksjfl/SEABO.

Pre-training on task-agnostic large datasets is a promising approach for enhancing the sample efficiency of reinforcement learning (RL) in solving complex tasks. We present PTGM, a novel method that pre-trains goal-based models to augment RL by providing temporal abstractions and behavior regularization. PTGM involves pre-training a low-level, goal-conditioned policy and training a high-level policy to generate goals for subsequent RL tasks. To address the challenges posed by the high-dimensional goal space, while simultaneously maintaining the agent's capability to accomplish various skills, we propose clustering goals in the dataset to form a discrete high-level action space. Additionally, we introduce a pre-trained goal prior model to regularize the behavior of the high-level policy in RL, enhancing sample efficiency and learning stability. Experimental results in a robotic simulation environment and the challenging open-world environment of Minecraft demonstrate PTGM’s superiority in sample efficiency and task performance compared to baselines. Moreover, PTGM exemplifies enhanced interpretability and generalization of the acquired low-level skills.

Recent studies have presented compelling evidence that large language models (LLMs) can equip embodied agents with the self-driven capability to interact with the world, which marks an initial step toward versatile robotics. However, these efforts tend to overlook the visual richness of open worlds, rendering the entire interactive process akin to ``a blindfolded text-based game.'' Consequently, LLM-based agents frequently encounter challenges in intuitively comprehending their surroundings and producing responses that are easy to understand. In this paper, we propose Steve-Eye, an end-to-end trained large multimodal model to address this limitation. Steve-Eye integrates the LLM with a visual encoder to process visual-text inputs and generate multimodal feedback. We adopt a semi-automatic strategy to collect an extensive dataset comprising 850K open-world instruction pairs, enabling our model to encompass three essential functions for an agent: multimodal perception, foundational knowledge base, and skill prediction and planning. Lastly, we develop three open-world evaluation benchmarks and carry out experiments from a wide range of perspectives to validate our model's capability to strategically act and plan. The project’s website and code can be found at https://sites.google.com/view/steve-eye.

In real-world scenarios, the application of reinforcement learning is significantly challenged by complex non-stationarity. Most existing methods attempt to model changes in the environment explicitly, often requiring impractical prior knowledge of environments. In this paper, we propose a new perspective, positing that non-stationarity can propagate and accumulate through complex causal relationships during state transitions, thereby compounding its sophistication and affecting policy learning. We believe that this challenge can be more effectively addressed by implicitly tracing the causal origin of non-stationarity. To this end, we introduce the Causal-Origin REPresentation (COREP) algorithm. COREP primarily employs a guided updating mechanism to learn a stable graph representation for the state, termed as causal-origin representation. By leveraging this representation, the learned policy exhibits impressive resilience to non-stationarity. We supplement our approach with a theoretical analysis grounded in the causal interpretation for non-stationary reinforcement learning, advocating for the validity of the causal-origin representation. Experimental results further demonstrate the superior performance of COREP over existing methods in tackling non-stationarity problems. The code is available at https://github.com/PKU-RL/COREP.

It is vital to learn effective policies that can be transferred to different domains with dynamics discrepancies in reinforcement learning (RL). In this paper, we consider dynamics adaptation settings where there exists dynamics mismatch between the source domain and the target domain, and one can get access to sufficient source domain data, while can only have limited interactions with the target domain. Existing methods address this problem by learning domain classifiers, performing data filtering from a value discrepancy perspective, etc. Instead, we tackle this challenge from a decoupled representation learning perspective. We perform representation learning only in the target domain and measure the representation deviations on the transitions from the source domain, which we show can be a signal of dynamics mismatch. We also show that representation deviation upper bounds performance difference of a given policy in the source domain and target domain, which motivates us to adopt representation deviation as a reward penalty. The produced representations are not involved in either policy or value function, but only serve as a reward penalizer. We conduct extensive experiments on environments with kinematic and morphology mismatch, and the results show that our method exhibits strong performance on many tasks. Our code is publicly available at https://github.com/dmksjfl/PAR.

The partial observability and stochasticity in multi-agent settings can be mitigated by accessing more information about others via communication. However, the coordination problem still exists since agents cannot communicate actual actions with each other at the same time due to the circular dependencies. In this paper, we propose a novel multi-level communication scheme, Sequential Communication (SeqComm). SeqComm treats agents asynchronously (the upper-level agents make decisions before the lower-level ones) and has two communication phases. In the negotiation phase, agents determine the priority of decision-making by communicating hidden states of observations and comparing the value of intention, which is obtained by modeling the environment dynamics. In the launching phase, the upper-level agents take the lead in making decisions and then communicate their actions with the lower-level agents. Theoretically, we prove the policies learned by SeqComm are guaranteed to improve monotonically and converge. Empirically, we show that SeqComm outperforms existing methods in a variety of cooperative multi-agent tasks.

Large Language Models (LLMs) have demonstrated proficiency in utilizing various tools by coding, yet they face limitations in handling intricate logic and precise control. In embodied tasks, high-level planning is amenable to direct coding, while low-level actions often necessitate task-specific refinement, such as Reinforcement Learning (RL). To seamlessly integrate both modalities, we introduce a two-level hierarchical framework, RL-GPT, comprising a slow agent and a fast agent. The slow agent analyzes actions suitable for coding, while the fast agent executes coding tasks. This decomposition effectively focuses each agent on specific tasks, proving highly efficient within our pipeline. Our approach outperforms traditional RL methods and existing GPT agents, demonstrating superior efficiency. In the Minecraft game, it rapidly obtains diamonds within a single day on an RTX3090. Additionally, it achieves SOTA performance across all designated MineDojo tasks.

We consider off-dynamics reinforcement learning (RL) where one needs to transfer policies across different domains with dynamics mismatch. Despite the focus on developing dynamics-aware algorithms, this field is hindered due to the lack of a standard benchmark. To bridge this gap, we introduce ODRL, the first benchmark tailored for evaluating off-dynamics RL methods. ODRL contains four experimental settings where the source and target domains can be either online or offline, and provides diverse tasks and a broad spectrum of dynamics shifts, making it a reliable platform to comprehensively evaluate the agent's adaptation ability to the target domain. Furthermore, ODRL includes recent off-dynamics RL algorithms in a unified framework and introduces some extra baselines for different settings, all implemented in a single-file manner. To unpack the true adaptation capability of existing methods, we conduct extensive benchmarking experiments, which show that no method has universal advantages across varied dynamics shifts. We hope this benchmark can serve as a cornerstone for future research endeavors. Our code is publicly available at https://github.com/OffDynamicsRL/off-dynamics-rl.

Decision-making in large-scale games is an essential research area in artificial intelligence (AI) with significant real-world impact. However, the limited access to realistic large-scale game environments has hindered research progress in this area. In this paper, we present AuctionNet, a benchmark for bid decision-making in large-scale ad auctions derived from a real-world online advertising platform. AuctionNet is composed of three parts: an ad auction environment, a pre-generated dataset based on the environment, and performance evaluations of several baseline bid decision-making algorithms. More specifically, the environment effectively replicates the integrity and complexity of real-world ad auctions through the interaction of several modules: the ad opportunity generation module employs deep generative networks to bridge the gap between simulated and real-world data while mitigating the risk of sensitive data exposure; the bidding module implements diverse auto-bidding agents trained with different decision-making algorithms; and the auction module is anchored in the classic Generalized Second Price (GSP) auction but also allows for customization of auction mechanisms as needed. To facilitate research and provide insights into the environment, we have also pre-generated a substantial dataset based on the environment. The dataset contains 10 million ad opportunities, 48 diverse auto-bidding agents, and over 500 million auction records. Performance evaluations of baseline algorithms such as linear programming, reinforcement learning, and generative models for bid decision-making are also presented as a part of AuctionNet. AuctionNet has powered the NeurIPS 2024 Auto-Bidding in Large-Scale Auctions competition, providing competition environments for over 1,500 teams. We believe that AuctionNet is applicable not only to research on bid decision-making in ad auctions but also to the general area of decision-making in large-scale games. Code: https://github.com/alimama-tech/AuctionNet.

When an agent is in a multi-agent environment, it may face previously unseen opponents, and it is a challenge to cooperate with other agents to accomplish the task together or to maximize its own rewards. Most opponent modeling methods deal with the non-stationarity caused by unknown opponent policies via predicting the opponent’s actions. However, focusing on the opponent’s action is shortsighted, which also constrains the adaptability to unknown opponents in complex tasks. In this paper, we propose opponent modeling based on subgoal inference, which infers the opponent’s subgoals through historical trajectories. As subgoals are likely to be shared by different opponent policies, predicting subgoals can yield better generalization to unknown opponents. Additionally, we design two subgoal selection modes for cooperative games and general-sum games respectively. Empirically, we show that our method achieves more effective adaptation than existing methods in a variety of tasks.

Efficiently solving unseen tasks remains a challenge in reinforcement learning (RL), especially for long-horizon tasks composed of multiple subtasks. Pre-training policies from task-agnostic datasets has emerged as a promising approach, yet existing methods still necessitate substantial interactions via RL to learn new tasks. We introduce MGPO, a method that leverages the power of Transformer-based policies to model sequences of goals, enabling efficient online adaptation through prompt optimization. In its pre-training phase, MGPO utilizes hindsight multi-goal relabeling and behavior cloning. This combination equips the policy to model diverse long-horizon behaviors that align with varying goal sequences. During online adaptation, the goal sequence, conceptualized as a prompt, is optimized to improve task performance. We adopt a multi-armed bandit framework for this process, enhancing prompt selection based on the returns from online trajectories. Our experiments across various environments demonstrate that MGPO holds substantial advantages in sample efficiency, online adaptation performance, robustness, and interpretability compared with existing methods.

Recently, there are many efforts attempting to learn useful policies for continuous control in visual reinforcement learning (RL). In this scenario, it is important to learn a generalizable policy, as the testing environment may differ from the training environment, e.g., there exist distractors during deployment. Many practical algorithms are proposed to handle this problem. However, to the best of our knowledge, none of them provide a theoretical understanding of what affects the generalization gap and why their proposed methods work. In this paper, we bridge this issue by theoretically answering the key factors that contribute to the generalization gap when the testing environment has distractors. Our theories indicate that minimizing the representation distance between training and testing environments, which aligns with human intuition, is the most critical for the benefit of reducing the generalization gap. Our theoretical results are supported by the empirical evidence in the DMControl Generalization Benchmark (DMC-GB).

Achieving human-level dexterity in robotics remains a critical open problem. Even simple dexterous manipulation tasks pose significant difficulties due to the high number of degrees of freedom and the need for cooperation among heterogeneous agents (e.g., finger joints). While some researchers have utilized reinforcement learning (RL) to control a single hand in manipulating objects, tasks that require coordinated bimanual cooperation are still under-explored due to the fewer suitable environments, which can result in difficulties and sub-optimal performance. To address these challenges, we introduce Bi-DexHands, a simulator with two dexterous hands featuring 20 bimanual manipulation tasks and thousands of target objects, designed to match various levels of human motor skills based on cognitive science research. We developed Bi-DexHands in Issac Gym, enabling highly efficient RL training at over 30,000 frames per second using a single NVIDIA RTX 3090. Based on Bi-DexHands, we present a comprehensive evaluation of popular RL algorithms in different settings, including single-agent/multi-agent RL, offline RL, multi-task RL, and meta RL. Our findings show that on-policy algorithms, such as PPO, can master simple manipulation tasks that correspond to those of 48-month-old babies, such as catching a flying object or opening a bottle. Furthermore, multi-agent RL can improve the ability to perform manipulations that require skilled bimanual cooperation, such as lifting a pot or stacking blocks. Despite achieving success in individual tasks, current RL algorithms struggle to learn multiple manipulation skills in most multi-task and few-shot learning scenarios. This highlights the need for further research and development within the RL community.

Offline reinforcement learning could learn effective policies from a fixed dataset, which is promising for real-world applications. However, in offline decentralized multi-agent reinforcement learning, due to the discrepancy between the behavior policy and learned policy, the transition dynamics in offline experiences do not accord with the transition dynamics in online execution, which creates severe errors in value estimates, leading to uncoordinated low-performing policies. One way to overcome this problem is to bridge offline training and online tuning. However, considering both deployment efficiency and sample efficiency, we could only collect very limited online experiences, making it insufficient to use merely online data for updating the agent policy. To utilize both offline and online experiences to tune the policies of agents, we introduce online transition correction (OTC) to implicitly correct the offline transition dynamics by modifying sampling probabilities. We design two types of distances, i.e., embedding-based and value-based distance, to measure the similarity between transitions, and further propose an adaptive rank-based prioritization to sample transitions according to the transition similarity. OTC is simple yet effective to increase data efficiency and improve agent policies in online tuning. Empirically, OTC outperforms baselines in a variety of tasks.

In this paper, we propose multi-agent automated machine learning (MA2ML) with the aim to effectively handle joint optimization of modules in automated machine learning (AutoML). MA2ML takes each machine learning module, such as data augmentation (AUG), neural architecture search (NAS), or hyper-parameters (HPO), as an agent and the final performance as the reward, to formulate a multi-agent reinforcement learning problem. MA2ML explicitly assigns credit to each agent according to its marginal contribution to enhance cooperation among modules, and incorporates off-policy learning to improve search efficiency. Theoretically, MA2ML guarantees monotonic improvement of joint optimization. Extensive experiments show that MA2ML yields the state-of-the-art top-1 accuracy on ImageNet under constraints of computational cost, e.g., 79.7%/80.5% with FLOPs fewer than 600M/800M. Exten\sive ablation studies verify the benefits of credit assignment and off-policy learning of MA2ML.

In many real-world multi-agent cooperative tasks, due to high cost and risk, agents cannot continuously interact with the environment and collect experiences during learning, but have to learn from offline datasets. However, the transition dynamics in the dataset of each agent can be much different from the ones induced by the learned policies of other agents in execution, creating large errors in value estimates. Consequently, agents learn uncoordinated low-performing policies. In this paper, we propose a framework for offline decentralized multi-agent reinforcement learning, which exploits value deviation and transition normalization to deliberately modify the transition probabilities. Value deviation optimistically increases the transition probabilities of high-value next states, and transition normalization normalizes the transition probabilities of next states. They together enable agents to learn high-performing and coordinated policies. Theoretically, we prove the convergence of Q-learning under the altered non-stationary transition dynamics. Empirically, we show that the framework can be easily built on many existing offline reinforcement learning algorithms and achieve substantial improvement in a variety of multi-agent tasks.

Illumination planning in photometric stereo aims to find a balance between surface normal estimation accuracy and image capturing efficiency by selecting optimal light configurations. It depends on factors such as the unknown shape and general reflectance of the target object, global illumination, and the choice of photometric stereo backbones, which are too complex to be handled by existing methods based on handcrafted illumination planning rules. This paper proposes a learning-based illumination planning method that jointly considers these factors via integrating a neural network and a generalized image formation model. As it is impractical to supervise illumination planning due to the enormous search space for ground truth light configurations, we formulate illumination planning using reinforcement learning, which explores the light space in a photometric stereo-aware and reward-driven manner. Experiments on synthetic and real-world datasets demonstrate that photometric stereo under the 20-light configurations from our method is comparable to, or even surpasses that of using lights from all available directions.

We investigate the use of natural language to drive the generalization of policies in multi-agent settings. Unlike single-agent settings, the generalization of policies should also consider the influence of other agents. Besides, with the increasing number of entities in multi-agent settings, more agent-entity interactions are needed for language grounding, and the enormous search space could impede the learning process. Moreover, given a simple general instruction, e.g., beating all enemies, agents are required to decompose it into multiple subgoals and figure out the right one to focus on. Inspired by previous work, we try to address these issues at the entity level and propose a novel framework for language grounding in multi-agent reinforcement learning, entity divider (EnDi). EnDi enables agents to independently learn subgoal division at the entity level and act in the environment based on the associated entities. The subgoal division is regularized by agent modeling to avoid subgoal conflicts and promote coordinated strategies. Empirically, EnDi demonstrates the strong generalization ability to unseen games with new dynamics and expresses the superiority over existing methods. The code is available at https://github.com/PKU-RL/EnDi.

Despite the success of cooperative multi-agent reinforcement learning algorithms, most of them focus on a single team composition, which prevents them from being used in more realistic scenarios where dynamic team composition is possible. While some studies attempt to solve this problem via multi-task learning in a fixed set of team compositions, there is still a risk of overfitting to the training set, which may lead to catastrophic performance when facing dramatically varying team compositions during execution. To address this problem, we propose to use mutual information (MI) as an augmented reward to prevent individual policies from relying too much on team-related information and encourage agents to learn policies that are robust in different team compositions. Optimizing this MI-augmented objective in an off-policy manner can be intractable due to the existence of dynamic marginal distribution. To alleviate this problem, we first propose a multi-agent policy iteration algorithm with a fixed marginal distribution and prove its convergence and optimality. Then, we propose to employ the Blahut–Arimoto algorithm and an imaginary team composition distribution for optimization with approximate marginal distribution as the practical implementation. Empirically, our method demonstrates strong zero-shot generalization to dynamic team compositions in complex cooperative tasks.

Learning from visual observation (LfVO), aiming at recovering policies from only visual observation data, is promising yet a challenging problem. Existing LfVO approaches either only adopt inefficient online learning schemes or require additional task-specific information like goal states, making them not suited for open-ended tasks. To address these issues, we propose a two-stage framework for learning from visual observation. In the first stage, we introduce and pretrain State-to-Go (STG) Transformer offline to predict and differentiate latent transitions of demonstrations. Subsequently, in the second stage, the STG Transformer provides intrinsic rewards for downstream reinforcement learning tasks where an agent learns merely from intrinsic rewards. Empirical results on Atari and Minecraft show that our proposed method outperforms baselines and in some tasks even achieves performance comparable to the policy learned from environmental rewards. These results shed light on the potential of utilizing video-only data to solve difficult visual reinforcement learning tasks rather than relying on complete offline datasets containing states, actions, and rewards. The project’s website and code can be found at https://sites.google.com/view/stgtransformer.

Traffic signal control aims to coordinate traffic signals across intersections to improve the traffic efficiency of a district or a city. Deep reinforcement learning (RL) has been applied to traffic signal control recently and demonstrated promising performance where each traffic signal is regarded as an agent. However, there are still several challenges that may limit its large-scale application in the real world. On the one hand, the policy of the current traffic signal is often heavily influenced by its neighbor agents, and the coordination between the agent and its neighbors needs to be considered. Hence, the control of a road network composed of multiple traffic signals is naturally modeled as a multi-agent system, and all agents’ policies need to be optimized simultaneously. On the other hand, once the policy function is conditioned on not only the current agent's observation but also the neighbors’, the policy function would be closely related to the training scenario and cause poor generalizability because the agents in various scenarios often have heterogeneous neighbors. To make the policy learned from a training scenario generalizable to new unseen scenarios, a novel Meta Variationally Intrinsic Motivated (MetaVIM) RL method is proposed to learn the decentralized policy for each intersection that considers neighbor information in a latent way. Specifically, we formulate the policy learning as a meta-learning problem over a set of related tasks, where each task corresponds to traffic signal control at an intersection whose neighbors are regarded as the unobserved part of the state. Then, a learned latent variable is introduced to represent the task's specific information and is further brought into the policy for learning. In addition, to make the policy learning stable, a novel intrinsic reward is designed to encourage each agent's received rewards and observation transition to be predictable only conditioned on its own history. Extensive experiments conducted on CityFlow demonstrate that the proposed method substantially outperforms existing approaches and shows superior generalizability.

Multi-agent policy gradient methods in centralized training with decentralized execution recently witnessed many progresses. During centralized training, multi-agent credit assignment is crucial, which can substantially promote learning performance. However, explicit multi-agent credit assignment in multi-agent policy gradient methods still receives less attention. In this paper, we investigate multi-agent credit assignment induced by reward shaping and provide a theoretical understanding in terms of its credit assignment and policy bias. Based on this, we propose an exponentially weighted advantage estimator, which is analogous to GAE, to enable multi-agent credit assignment while allowing the tradeoff with policy bias. Empirical results show that our approach can successfully perform effective multi-agent credit assignment, and thus substantially outperforms other advantage estimators.

Entropy regularization is a popular method in reinforcement learning (RL). Although it has many advantages, it alters the RL objective and makes the converged policy deviate from the optimal policy of the original Markov Decision Process (MDP). Though divergence regularization has been proposed to settle this problem, it cannot be trivially applied to cooperative multi-agent reinforcement learning (MARL). In this paper, we investigate divergence regularization in cooperative MARL and propose a novel off-policy cooperative MARL framework, divergence-regularized multi-agent actor-critic (DMAC). Theoretically, we derive the update rule of DMAC which is naturally off-policy, guarantees the monotonic policy improvement and convergence in both the original MDP and the divergence-regularized MDP, and is not biased by the regularization. We also give a bound of the discrepancy between the converged policy and the optimal policy in the original MDP. DMAC is a flexible framework and can be combined with many existing MARL algorithms. Empirically, we evaluate DMAC in a didactic stochastic game and StarCraft Multi-Agent Challenge and show that DMAC substantially improves the performance of existing MARL algorithms.

We study offline meta-reinforcement learning, a practical reinforcement learning paradigm that learns from offline data to adapt to new tasks. The distribution of offline data is determined jointly by the behavior policy and the task. Existing offline meta-reinforcement learning algorithms cannot distinguish these factors, making task representations unstable to the change of behavior policies. To address this problem, we propose a contrastive learning framework for task representations that are robust to the distribution mismatch of behavior policies in training and test. We design a bi-level encoder structure, use mutual information maximization to formalize task representation learning, derive a contrastive learning objective, and introduce several approaches to approximate the true distribution of negative pairs. Experiments on a variety of offline meta-reinforcement learning benchmarks demonstrate the advantages of our method over prior methods, especially on the generalization to out-of-distribution behavior policies.

Achieving human-level dexterity is an important open problem in robotics. However, tasks of dexterous hand manipulation even at the baby level are challenging to solve through reinforcement learning (RL). The difficulty lies in the high degrees of freedom and the required cooperation among heterogeneous agents (e.g., joints of fingers). In this study, we propose the Bimanual Dexterous Hands Benchmark (Bi-DexHands), a simulator that involves two dexterous hands with tens of bimanual manipulation tasks and thousands of target objects. Tasks in Bi-DexHands are first designed to match human-level motor skills according to literature in cognitive science, and then are built in Issac Gym; this enables highly efficient RL trainings, reaching 30,000+ FPS by only one single NVIDIA RTX 3090. We provide a comprehensive benchmark for popular RL algorithms under different settings; this includes multi-agent RL, offline RL, multi-task RL, and meta RL. Our results show that PPO type on-policy algorithms can learn to solve simple manipulation tasks that are equivalent up to 48-month human baby (e.g., catching a flying object, opening a bottle), while multi-agent RL can further help to learn manipulations that require skilled bimanual cooperation (e.g., lifting a pot, stacking blocks). Despite the success on each individual task, when it comes to mastering multiple manipulation skills, existing RL algorithms fail to work in most of the multi-task and the few-shot learning tasks, which calls for more future development from the RL community. Our project is open-sourced at https://github.com/PKU-MARL/DexterousHands.

Fully decentralized multi-agent reinforcement learning has shown great potentials for many real-world cooperative tasks, where the global information, \textit{e.g.}, the actions of other agents, is not accessible. Although independent Q-learning is widely used for decentralized training, the transition probabilities are non-stationary since other agents are updating policies simultaneously, which leads to non-guaranteed convergence of independent Q-learning. To deal with non-stationarity, we first introduce stationary ideal transition probabilities, on which independent Q-learning could converge to the global optimum. Further, we propose a fully decentralized method, I2Q, which performs independent Q-learning on the modeled ideal transition function to reach the global optimum. The modeling of ideal transition function in I2Q is fully decentralized and independent from the learned policies of other agents, helping I2Q be free from non-stationarity and learn the optimal policy. Empirically, we show that I2Q can achieve remarkable improvement in a variety of cooperative multi-agent tasks.

The learned policy of model-free offline reinforcement learning (RL) methods is often constrained to stay within the support of datasets to avoid possible dangerous out-of-distribution actions or states, making it challenging to handle out-of-support region. Model-based RL methods offer a richer dataset and benefit generalization by generating imaginary trajectories with either trained forward or reverse dynamics model. However, the imagined transitions may be inaccurate, thus downgrading the performance of the underlying offline RL method. In this paper, we propose to augment the offline dataset by using trained bidirectional dynamics models and rollout policies with double check. We introduce conservatism by trusting samples that the forward model and backward model agree on. Our method, confidence-aware bidirectional offline model-based imagination, generates reliable samples and can be combined with any model-free offline RL method. Experimental results on the D4RL benchmarks demonstrate that our method significantly boosts the performance of existing model-free offline RL algorithms and achieves competitive or better scores against baseline methods.

Offline reinforcement learning (RL) defines the task of learning from a static logged dataset without continually interacting with the environment. The distribution shift between the learned policy and the behavior policy makes it necessary for the value function to stay conservative such that out-of-distribution (OOD) actions will not be severely overestimated. However, existing approaches, penalizing the unseen actions or regularizing with the behavior policy, are too pessimistic, which suppresses the generalization of the value function and hinders the performance improvement. This paper explores mild but enough conservatism for offline learning while not harming generalization. We propose Mildly Conservative Q-learning (MCQ), where OOD actions are actively trained by assigning them proper pseudo Q values. We theoretically show that MCQ induces a policy that behaves at least as well as the behavior policy and no erroneous overestimation will occur for OOD actions. Experimental results on the D4RL benchmarks demonstrate that MCQ achieves remarkable performance compared with prior work. Furthermore, MCQ shows superior generalization ability when transferring from offline to online, and significantly outperforms baselines. Our code is publicly available at https://github.com/dmksjfl/MCQ.

In this paper, we study the problem of networked multi-agent reinforcement learning (MARL), where a number of agents are deployed as a partially connected network and each interacts only with nearby agents. Networked MARL requires all agents to make decisions in a decentralized manner to optimize a global objective with restricted communication between neighbors over the network. Inspired by the fact that sharing plays a key role in human's learning of cooperation, we propose LToS, a hierarchically decentralized MARL framework that enables agents to learn to dynamically share reward with neighbors so as to encourage agents to cooperate on the global objective through collectives. For each agent, the high-level policy learns how to share reward with neighbors to decompose the global objective, while the low-level policy learns to optimize the local objective induced by the high-level policies in the neighborhood. The two policies form a bi-level optimization and learn alternately. We empirically demonstrate that LToS outperforms existing methods in both social dilemma and networked MARL scenarios across scales.

When one agent interacts with a multi-agent environment, it is challenging to deal with various opponents unseen before. Modeling the behaviors, goals, or beliefs of opponents could help the agent adjust its policy to adapt to different opponents. In addition, it is also important to consider opponents who are learning simultaneously or capable of reasoning. However, existing work usually tackles only one of the aforementioned types of opponents. In this paper, we propose model-based opponent modeling (MBOM), which employs the environment model to adapt to all kinds of opponents. MBOM simulates the recursive reasoning process in the environment model and imagines a set of improving opponent policies. To effectively and accurately represent the opponent policy, MBOM further mixes the imagined opponent policies according to the similarity with the real behaviors of opponents. Empirically, we show that MBOM achieves more effective adaptation than existing methods in a variety of tasks, respectively with different types of opponents, i.e., fixed policy, naive learner, and reasoning learner.

The net orders in detailed routing are crucial to routing closure, especially in most modern routers following the sequential routing manner with the rip-up and reroute scheme. In advanced technology nodes, detailed routing has to deal with complicated design rules and large problem sizes, making its performance more sensitive to the order of nets to be routed. In the literature, the net orders are mostly determined by simple heuristic rules tuned for specific benchmarks. In this work, we propose an asynchronous reinforcement learning (RL) framework to automatically search for optimal ordering strategies and a transfer learning (TL) algorithm to improve performance. By asynchronous querying, the router, pretraining the RL agents, and finetuning with the TL algorithm, we can generate high-performance routing sequences to achieve a 26% reduction in the DRC violations and a 1.2% reduction in the total costs compared with the state-of-the-art detailed router.

Deep reinforcement learning (RL) has been applied to traffic signal control recently and demonstrated superior performance to conventional control methods. However, there are still several challenges we have to address before fully applying deep RL to traffic signal control. Firstly, the objective of traffic signal control is to optimize average travel time, which is a delayed reward in a long time horizon in the context of RL. However, existing work simplifies the optimization by using queue length, waiting time, delay, etc., as immediate reward and presumes these short-term targets are always aligned with the objective. Nevertheless, these targets may deviate from the objective in different road networks with various traffic patterns. Secondly, it remains unsolved how to cooperatively control traffic signals to directly optimize average travel time. To address these challenges, we propose a hierarchical and cooperative reinforcement learning method-HiLight. HiLight enables each agent to learn a high-level policy that optimizes the objective locally by selecting among the sub-policies that respectively optimize short-term targets. Moreover, the high-level policy additionally considers the objective in the neighborhood with adaptive weighting to encourage agents to cooperate on the objective in the road network. Empirically, we demonstrate that HiLight outperforms state-of-the-art RL methods for traffic signal control in real road networks with real traffic.

The net orders in detailed routing are crucial to routing closure, especially in most modern routers following the sequential routing manner with the rip-up and reroute scheme. In advanced technology nodes, detailed routing has to deal with complicated design rules and large problem sizes, making its performance more sensitive to the order of nets to be routed. In literature, the net orders are mostly determined by simple heuristic rules tuned for specific benchmarks. In this work, we propose an asynchronous reinforcement learning (RL) framework to search for optimal ordering strategies automatically. By asynchronous querying the router and training the RL agents, we can generate highperformance routing sequences to achieve better solution quality.

Individuality is essential in human society. It induces the division of labor and thus improves the efficiency and productivity. Similarly, it should also be a key to multi-agent cooperation. Inspired by that individuality is of being an individual separate from others, we propose a simple yet efficient method for the emergence of individuality (EOI) in multi-agent reinforcement learning (MARL). EOI learns a probabilistic classifier that predicts a probability distribution over agents given their observation and gives each agent an intrinsic reward of being correctly predicted by the classifier. The intrinsic reward encourages the agents to visit their own familiar observations, and learning the classifier by such observations makes the intrinsic reward signals stronger and in turn makes the agents more identifiable. To further enhance the intrinsic reward and promote the emergence of individuality, two regularizers are proposed to increase the discriminability of the classifier. We implement EOI on top of popular MARL algorithms. Empirically, we show that EOI outperforms existing methods in a variety of multi-agent cooperative scenarios.

Value decomposition recently injects vigorous vitality into multi-agent actor-critic methods. However, existing decomposed actor-critic methods cannot guarantee the convergence of global optimum. In this paper, we present a novel multi-agent actor-critic method, FOP, which can factorize the optimal joint policy induced by maximum-entropy multi-agent reinforcement learning (MARL) into individual policies. Theoretically, we prove that factorized individual policies of FOP converge to the global optimum. Empirically, in the well-known matrix game and differential game, we verify that FOP can converge to the global optimum for both discrete and continuous action spaces. We also evaluate FOP on a set of StarCraft II micromanagement tasks, and demonstrate that FOP substantially outperforms state-of-the-art decomposed value-based and actor-critic methods.

Mobile devices collect a large amount of visual data that are useful for many applications. Searching for an object of interest over a network of mobile devices can aid human analysts in a variety of situations. However, processing the information on these devices is a challenge owing to the high computational complexity of the state-of-the-art computer vision algorithms that primarily rely on Convolutional Neural Networks (CNNs). Thus, this paper builds PicSys, a system that enables answering visual search queries on a mobile network. The objective of the system is to minimize the maximum completion time over all devices while taking into account the energy consumption of mobile devices as well. First, PicSys carefully divides the computation into multiple filtering stages, such that only a small percentage of images need to run the entire CNN pipeline. Splitting such CNN computation into multiple stages requires understanding the intermediate CNN features and systematically trading off accuracy for the computation speed. Second, PicSys determines where to run each of the stages of the multi-stage pipeline to fully utilize the available resources. Finally, through extensive experimentation, system implementation, and simulation, we show that PicSys performance is close to optimal and significantly outperforms other standard algorithms.

Convolutional Neural Networks (ConvNets/CNNs) have revolutionized the research in computer vision, due to their ability to capture complex patterns, resulting in high inference accuracies. However, the increasingly complex nature of these neural networks means that they are particularly suited for server computers with powerful GPUs. We envision that deep learning applications will be eventually widely deployed on mobile devices, e.g., smartphones, self-driving cars, and drones. Therefore, in this paper, we aim to understand the resource requirements of CNNs on mobile devices in terms of compute time, memory, and power. First, by deploying several popular CNNs on different mobile CPUs and GPUs, we measure and analyze the performance and resource usage for the CNNs on a layerwise granularity. Our findings point out the potential ways of optimizing the CNN pipelines on mobile devices. Second, we model resource requirements of core computations of CNNs. Finally, based on the measurement and modeling, we build and evaluate our modeling tool, Augur, which takes a CNN configuration (descriptor) as the input and estimates the compute time, memory, and power requirements of the CNN, to give insights about whether and how efficiently a CNN can be run on a given mobile platform.

Sparse reward is one of the biggest challenges in reinforcement learning (RL). In this paper, we propose a novel method called Generative Exploration and Exploitation (GENE) to overcome sparse reward. GENE automatically generates start states to encourage the agent to explore the environment and to exploit received reward signals. GENE can adaptively tradeoff between exploration and exploitation according to the varying distributions of states experienced by the agent as the learning progresses. GENE relies on no prior knowledge about the environment and can be combined with any RL algorithm, no matter on-policy or off-policy, single-agent or multi-agent. Empirically, we demonstrate that GENE significantly outperforms existing methods in three tasks with only binary rewards, including Maze, Maze Ant, and Cooperative Navigation. Ablation studies verify the emergence of progressive exploration and automatic reversing.

Communication lays the foundation for human cooperation. It is also crucial for multi-agent cooperation. However, existing work focuses on broadcast communication, which is not only impractical but also leads to information redundancy that could even impair the learning process. To tackle these difficulties, we propose Individually Inferred Communication (I2C), a simple yet effective model to enable agents to learn a prior for agent-agent communication. The prior knowledge is learned via causal inference and realized by a feed-forward neural network that maps the agent's local observation to a belief about who to communicate with. The influence of one agent on another is inferred via the joint action-value function in multi-agent reinforcement learning and quantified to label the necessity of agent-agent communication. Furthermore, the agent policy is regularized to better exploit communicated messages. Empirically, we show that I2C can not only reduce communication overhead but also improve the performance in a variety of multi-agent cooperative scenarios, comparing to existing methods.

The vast adoption of mobile devices with cameras has greatly contributed to the proliferation of the creation and distribution of videos. For a variety of purposes, valuable information may be extracted from these videos. While the computational capability of mobile devices has greatly improved recently, video processing is still a demanding task for mobile devices. We design an on-demand video processing system, NetVision, that performs distributed video processing using deep learning across a wireless network of mobile and edge devices to answer queries while minimizing the query response time. However, the problem of minimal query response time for processing videos stored across a network is a strongly NP-hard problem. To deal with this, we design a greedy algorithm with bounded performance. To further deal with the dynamics of the transmission rate between mobile and edge devices, we design an adaptive algorithm. We built NetVision and deployed it on a small testbed. Based on the measurements of the testbed and by extensive simulations, we show that the greedy algorithm is close to the optimum and the adaptive algorithm performs better with more dynamic transmission rates. We then perform experiments on the small testbed to examine the realized system performance in both stationary networks and mobile networks.

Fairness is essential for human society, contributing to stability and productivity. Similarly, fairness is also the key for many multi-agent systems. Taking fairness into multi-agent learning could help multi-agent systems become both efficient and stable. However, learning efficiency and fairness simultaneously is a complex, multi-objective, joint-policy optimization. To tackle these difficulties, we propose FEN, a novel hierarchical reinforcement learning model. We first decompose fairness for each agent and propose fair-efficient reward that each agent learns its own policy to optimize. To avoid multi-objective conflict, we design a hierarchy consisting of a controller and several sub-policies, where the controller maximizes the fair-efficient reward by switching among the sub-policies that provides diverse behaviors to interact with the environment. FEN can be trained in a fully decentralized way, making it easy to be deployed in real-world applications. Empirically, we show that FEN easily learns both fairness and efficiency and significantly outperforms baselines in a variety of multi-agent scenarios.

Mobile devices such as smartphones are enabling users to generate and share videos with increasing rates. In some cases, these videos may contain valuable information, which can be exploited for a variety of purposes. However, instead of centrally collecting and processing videos for information retrieval, we consider crowdprocessing videos, where each mobile device locally processes stored videos. While the computational capability of mobile devices continues to improve, processing videos using deep learning, i.e., convolutional neural networks, is still a demanding task for mobile devices. To this end, we design and build CrowdVision, a computing platform that enables mobile devices to crowdprocess videos using deep learning in a distributed and energy-efficient manner leveraging cloud offload. CrowdVision can quickly and efficiently process videos with offload under various settings and different network connections and greatly outperform the existing computation offload framework (e.g., with a 2× speed-up). In doing so, CrowdVision tackles several challenges: (i) how to exploit the characteristics of the computing of deep learning for video processing; (ii) how to parallelize processing and offloading for acceleration; and (iii) how to optimize both time and energy at runtime by just determining the right moments to offload.

Mobile devices such as smartphones are enabling users to generate and share videos with increasing rates. In some cases, these videos may contain valuable information, which can be exploited for a variety of purposes. However, instead of centrally collecting and processing videos for information retrieval, we consider crowdprocessing videos, where each mobile device locally processes stored videos. While the computational capability of mobile devices continues to improve, processing videos using deep learning, i.e., convolutional neural networks, is still a demanding task for mobile devices. To this end, we design and build CrowdVision, a computing platform that enables mobile devices to crowdprocess videos using deep learning in a distributed and energy-efficient manner leveraging cloud offload. CrowdVision can quickly and efficiently process videos with offload under various settings and different network connections and greatly outperform the existing computation offload framework (e.g., with a 2× speed-up). In doing so CrowdVision tackles several challenges: (i) how to exploit the characteristics of the computing of deep learning for video processing; (ii) how to parallelize processing and offloading for acceleration; and (iii) how to optimize both time and energy at runtime by just determining the right moments to offload.

Communication could potentially be an effective way for multi-agent cooperation. However, information sharing among all agents or in predefined communication architectures that existing methods adopt can be problematic. When there is a large number of agents, agents cannot differentiate valuable information that helps cooperative decision making from globally shared information. Therefore, communication barely helps, and could even impair the learning of multi-agent cooperation. Predefined communication architectures, on the other hand, restrict communication among agents and thus restrain potential cooperation. To tackle these difficulties, in this paper, we propose an attentional communication model that learns when communication is needed and how to integrate shared information for cooperative decision making. Our model leads to efficient and effective communication for large-scale multi-agent cooperation. Empirically, we show the strength of our model in a variety of cooperative scenarios, where agents are able to develop more coordinated and sophisticated strategies than existing methods.

Convolutional Neural Networks (CNNs) have revolutionized the research in computer vision, due to their ability to capture complex patterns, resulting in high inference accuracies. However, the increasingly complex nature of these neural networks means that they are particularly suited for server computers with powerful GPUs. We envision that deep learning applications will be eventually and widely deployed on mobile devices, e.g., smartphones, self-driving cars, and drones. Therefore, in this paper, we aim to understand the resource requirements (time, memory) of CNNs on mobile devices. First, by deploying several popular CNNs on mobile CPUs and GPUs, we measure and analyze the performance and resource usage for every layer of the CNNs. Our findings point out the potential ways of optimizing the performance on mobile devices. Second, we model the resource requirements of the different CNN computations. Finally, based on the measurement, profiling, and modeling, we build and evaluate our modeling tool, Augur, which takes a CNN configuration (descriptor) as the input and estimates the compute time and resource usage of the CNN, to give insights about whether and how efficiently a CNN can be run on a given mobile platform. In doing so Augur tackles several challenges: (i) how to overcome profiling and measurement overhead; (ii) how to capture the variance in different mobile platforms with different processors, memory, and cache sizes; and (iii) how to account for the variance in the number, type and size of layers of the different CNN configurations.

In this paper, we investigate how to network smartphones for providing communications in disaster recovery. By bridging the gaps among different kinds of wireless networks, we have designed and implemented a system called TeamPhone, which provides smartphones the capabilities of communications in disaster recovery. Specifically, TeamPhone consists of two components: A messaging system and a self-rescue system. The messaging system integrates cellular networking, ad-hoc networking, and opportunistic networking seamlessly, and enables communications among rescue workers. The self-rescue system groups, schedules, and positions the smartphones of trapped survivors. Such a group of smartphones can cooperatively wake up and send out emergency messages in an energy-efficient manner with their location and position information so as to assist rescue operations. We have implemented TeamPhone as a prototype application on the Android platform and deployed it on off-the-shelf smartphones. Experimental results demonstrate that TeamPhone can properly fulfill communication requirements and greatly facilitate rescue operations in disaster recovery.

Opportunistic mobile networks consisting of intermittently connected mobile devices have been exploited for various applications, such as computational offloading and mitigating cellular traffic load. In contrast to existing work, in this paper, we focus on cooperatively offloading data among mobile devices to maximally improve the probability of data delivery from a mobile device to intermittently connected infrastructure within a given time constraint, which is referred to as the cooperative offloading problem. Unfortunately, the estimation of data delivery probability over an opportunistic path is difficult and cooperative offloading is NP-hard. To this end, we first propose a probabilistic framework that provides the estimation of such probability. Based on the proposed probabilistic framework, we design a heuristic algorithm to solve cooperative offloading at a low computation cost. Due to the lack of global information, a distributed algorithm is further proposed. The performance of the proposed approaches is evaluated based on both synthetic networks and real traces. Experimental results show that the probabilistic framework can accurately estimate the data delivery probability, cooperative offloading greatly improves the delivery probability, the heuristic algorithm approximates the optimum, and the performance of both the heuristic algorithm and distributed algorithm outperforms other approaches.

Media cloud has become a promising paradigm for deploying large-scale streaming media applications at a reduced cost. Due to dynamic and diverse demands of users, media cloud presents two crucial characteristics: high resource consumption and dynamic traffic among media servers. Consequently, Virtual Machine (VM) migration in media cloud is highly required to suit varying resource requirements and the dynamic traffic patterns. Moreover, migration of such bandwidth-intensive media applications in media cloud needs cautious handling, especially for the internal traffic of Data Center Networks (DCN). However, existing media cloud resource management schemes or traffic-aware VM deployment approaches are insufficient for media cloud, ignoring the characteristics of either cloud infrastructure or media streaming requirements. In this paper, we propose a cluster-aware VM collaborative migration scheme for media cloud, tightly integrating clustering, placement, and dynamic migration process. The scheme employs a clustering algorithm and a placement algorithm to obtain ideal migration strategies for newly perceived media server clusters, and a migration algorithm to effectively accomplish the migration process of media servers. Evaluation results demonstrate that our scheme can effectively migrate virtual media servers in media cloud, while reducing the total internal traffic in DCN under the resource consumption constraints of media streaming applications.

The vast adoption of mobile devices with cameras has greatly assisted in the proliferation of the creation and distribution of videos. For a variety of purposes, valuable information may be extracted from these videos. While the computational capability of mobile devices has greatly improved recently, video processing is still a demanding task for mobile devices. Given a network consisting of mobile devices and video-clouds, mobile devices may be able to upload videos to video-clouds, which are more computationally capable for these processing tasks. However, due to networking constraints, when a video processing task is initiated through a query, most videos will not likely have been uploaded to the video-clouds, especially when the query is about a recent event. We investigate the problem of minimal query response time for processing videos stored across a network; however, this problem is a strongly NP-hard problem. To deal with this, we first propose a greedy algorithm with bounded performance. To further deal with the dynamics of the transmission rate between mobile devices and video-clouds, we propose an adaptive algorithm. To evaluate these algorithms, we built an on-demand video processing system. Based on the measurements of the system, we perform simulations to extensively evaluate the proposed algorithms. We also perform experiments on a small testbed to examine the realized system performance. Results show the performance of the greedy algorithm is close to the optimal and much better than other approaches, and the adaptive algorithm performs better with more dynamic transmission rates.

Opportunistic mobile networks consisting of intermittently connected mobile devices have been exploited for various applications, such as computational offloading and mitigating cellular traffic load. Different from existing work, in this paper, we focus on cooperatively offloading data among mobile devices to maximally improve the probability of data delivery from a mobile device to an intermittently connected remote server or data center within a given time constraint, which is referred to as the cooperative offloading problem. Unfortunately, cooperative offloading is NP-hard. To this end, a heuristic algorithm is designed based on the proposed probabilistic framework, which provides the estimation of the probability of successful data delivery over the opportunistic path, considering both data size and contact duration. Due to the lack of global information, a distributed algorithm is further proposed. The performance of the proposed approaches is evaluated based on both synthetic networks and real traces, and simulation results show that cooperative offloading can significantly improve the data delivery probability and the performance of both heuristic algorithm and distributed algorithm outperforms other approaches.

In this paper, we investigate how to network smart-phones for providing communications in disaster recovery. By bridging the gaps among different kinds of wireless networks, we have designed and implemented a system called TeamPhone, which provides smartphones the capabilities of communications in disaster recovery. Specifically, TeamPhone consists of two components: a messaging system and a self-rescue system. The messaging system integrates cellular networking, ad-hoc networking and opportunistic networking seamlessly, and enables communications among rescue workers. The self-rescue system energy-efficiently groups the smartphones of trapped survivor and sends out emergency messages so as to assist rescue operations. We have implemented TeamPhone as a prototype application on the Android platform and deployed it on off-the-shelf smartphones. Experiment results show that TeamPhone can properly fulfill communication requirements and greatly facilitate rescue operations in disaster recovery.

The emerging of mobile social networks opens opportunities for viral marketing. However, before fully utilizing mobile social networks as a platform for viral marketing, many challenges have to be addressed. In this paper, we address the problem of identifying a small number of individuals through whom the information can be diffused to the network as soon as possible, referred to as thediffusion minimizationproblem. Diffusion minimization under the probabilistic diffusion model can be formulated as an asymmetric$k$-center problem which is NP-hard, and the best known approximation algorithm for the asymmetric$k$-center problem has approximation ratio of$\log ^*n$and time complexity$O(n^5)$. Clearly, the performance and the time complexity of the approximation algorithm are not satisfiable in large-scale mobile social networks. To deal with this problem, we propose a community based algorithm and a distributed set-cover algorithm. The performance of the proposed algorithms is evaluated by extensive experiments on both synthetic networks and a real trace. The results show that the community based algorithm has the best performance in both synthetic networks and the real trace compared to existing algorithms, and the distributed set-cover algorithm outperforms the approximation algorithm in the real trace in terms of diffusion time.

This paper proposes SymDetector, a smartphone based application to unobtrusively detect the sound-related respiratory symptoms occurred in a user's daily life, including sneeze, cough, sniffle and throat clearing. SymDetector uses the built-in microphone on the smartphone to continuously monitor a user's acoustic data and uses multi-level processes to detect and classify the respiratory symptoms. Several practical issues are considered in developing SymDetector, such as users' privacy concerns about their acoustic data, resource constraints of the smartphone and different contexts of the smartphone. We have implemented SymDetector on Galaxy S3 and evaluated its performance in real experiments involving 16 users and 204 days. The experimental results show that SymDetector can detect these four types of respiratory symptoms with high accuracy under various conditions.

The ubiquity of mobile devices creates a rapidly growing market for mobile applications. Many of these applications involve complex processing tasks that are difficult to run on resource constrained mobile devices. This leads to the emergence of mobile cloud computing, in which cloud-based resources are used to enhance the computing capabilities of mobile devices. In this paper, we consider heterogeneous wireless networks in which multiple resource-rich computing nodes can be used as mobile clouds, and mobile devices can upload computation extensive tasks to these mobile clouds. The goal is to minimize the average task response time through determining whether to upload a task, and to which cloud the task should be uploaded. We formalize this task allocation problem, which is proved to be a NP-hard problem, and propose both offline centralized approach and online distributed approach to address this problem. Simulation results show that our approaches outperform others in terms of task response time in various scenarios.

Vaccination is one of the most effective ways to protect people from being infected by infectious disease. However, it is often impractical to vaccinate all people in a community due to various resource constraints. Therefore, targeted vaccination, which vaccinates a small group of people, is an alternative approach to contain infectious disease spread. To achieve better performance in targeted vaccination, we collect student contact traces in a high school based on wireless sensors carried by students. With our wireless sensor system, we can record student contacts within the disease propagation distance, and then construct a disease propagation graph to model the infectious disease propagation. Based on this graph, we propose a metric called connectivity centrality to measure a node's importance during disease propagation and design centrality based algorithms for targeted vaccination. The proposed algorithms are evaluated and compared with other schemes based on our collected traces. Trace driven simulation results show that our algorithms can help to effectively contain infectious disease.

Community detection is an important issue due to its wide use in designing network protocols such as data forwarding in Delay Tolerant Networks (DTN) and worm containment in Online Social Networks (OSN). However, most of the existing community detection algorithms focus on binary networks. Since most networks are naturally weighted such as DTN or OSN, in this article, we address the problems of community detection in weighted networks, exploit community for data forwarding in DTN and worm containment in OSN, and demonstrate how community can facilitate these network designs. Specifically, we propose a novel community detection algorithm, and introduce two metrics: intra-centrality and inter-centrality, to characterize nodes in communities, based on which we propose an efficient data forwarding algorithm for DTN and a worm containment strategy for OSN. Extensive trace-driven simulation results show that the proposed community detection algorithm, the data forwarding algorithm, and the worm containment strategy significantly outperform existing works.

The emerging of mobile social networks opens opportunities for viral marketing. However, before fully utilizing mobile social networks as a platform for viral marketing, many challenges have to be addressed. In this paper, we address the problem of identifying a small number of individuals through whom the information can be diffused to the network as soon as possible, referred to as the diffusion minimization problem. Diffusion minimization under the probabilistic diffusion model can be formulated as an asymmetric k-center problem which is NP-hard, and the best known approximation algorithm for the asymmetric k-center problem has approximation ratio of log*n and time complexity O(n5). Clearly, the performance and the time complexity of the approximation algorithm are not satisfiable in large-scale mobile social networks. To deal with this problem, we propose a community based algorithm and a distributed set-cover algorithm. The performance of the proposed algorithms is evaluated by extensive experiments on both synthetic networks and a real trace. The results show that the community based algorithm has the best performance in both synthetic networks and the real trace, and the distributed setcover algorithm outperforms the approximation algorithm in the real trace in terms of diffusion time.

Mobile social networks have emerged as a new frontier in the mobile computing research society, and the commonly used social structure (i.e., community) has been exploited to facilitate the design of network protocols and applications, such as data forwarding and worm containment. However, community based approaches may not be accurate when applied for predicting node contacts and may separate two frequently contacted nodes into different communities. In this paper, to address these problems, we propose skeleton, a tree structure specially designed for organizing network nodes, as the underlying structure in mobile social networks. We address the challenges on how to uncover skeleton from network, how to adapt skeleton with dynamic network and how to leverage skeleton for network protocol designs. Skeleton is constructed based on best friendship and skeleton construction is simple and efficient (e.g., less computational complexity than community detection). Algorithms are also designed to adapt skeleton construction to dynamic network. Moreover, a data forwarding algorithm and a worm containment strategy are designed based on skeleton. Trace-driven simulation results show that the skeleton based data forwarding algorithm and worm containment strategy outperform existing schemes based on community.

Community detection is an important issue due to its wide use in designing network protocols such as data forwarding in Delay Tolerant Networks (DTN) and worm containment in Online Social Networks (OSN). However, most of the existing community detection algorithms focus on binary networks. Since most networks are weighted such as social networks, DTN or OSN, in this paper, we address the problems of community detection in weighted networks and exploit community for data forwarding in DTN and worm containment in OSN. We propose a novel community detection algorithm, and then introduce two metrics called intra-centrality and inter-centrality, to characterize nodes in communities. Based on these metrics, we propose an efficient data forwarding algorithm for DTN and an efficient worm containment strategy for OSN. Extensive trace-driven simulation results show that the data forwarding algorithm and the worm containment strategy significantly outperform existing works.

In-network processing is touted as a key technology to eliminate data redundancy and minimize data transmission, which are crucial to saving energy in wireless sensor networks (WSNs). Specifically, operators participating in in-network processing are mapped to nodes in a sensor network. They receive data from downstream operators, process them and route the output to either the upstream operator or the sink node. The objective of operator tree placement is to minimize the total energy consumed in performing in-network processing. Two types of placement algorithms, centralized and distributed, have been proposed. A problem with the centralized algorithm is that it does not scale to large WSN's, because each sensor node is required to know the complete topology of the network. A problem with the distributed algorithm is their high message complexity. In this paper, we propose a heuristic algorithm to place a treestructured operator graph, and present a distributed implementation to optimize in-network processing cost and reduce the communication overhead. We prove a tight upper bound on the minimum in-network processing cost, and show that the heuristic algorithm has better performance than a canonical greedy algorithm. Simulation-based evaluations demonstrate the superior performance of our heuristic algorithm. We also give an improved distributed implementation of our algorithm that has a message overhead of O(M) per node, which is much less than the O(√NM log2M) and O(√NM) complexities for two previously proposed algorithms, Sync and MCFA, respectively. Here, N is the number of network nodes and M is the size of the operator tree.

Source-location privacy became one of major issues due to the open nature of wireless sensor networks. The adversary can eavesdrop and trace the message movements so as to capture the source. In the paper, first we propose Credit routing to provide the source-location privacy protection. Credit routing is able to route the message within the assigned credit at each message and randomize the routing path. Unlike other location privacy protection schemes in WSN, Credit routing not only can provide strong protection but also precisely control the transmission cost of each message. Then, we propose Hybrid credit routing, which routes the message to the receiver through three phases: totally random walk, forwarding random walk and credit random walk. These phases provide tri-fold protection to prevent the source from being captured by the adversary. We evaluate our proposed schemes based on several metrics including safe period, latency and protection efficiency. The simulation results show that Credit is able to provide the strong and efficient protection compared with other schemes including Phantom, LPR and RRIN. It is also shown Hybrid improves the protection strength and efficiency even further. The performance of Credit and Hybrid can be tuned by the assigned credit. For real application, the credit can be the real power consumption for forwarding the message from the source to the sink. So both Credit and Hybrid can be used to precisely control the power consumption for source-location protection.

Due to energy limitation of wireless sensor networks, in-network aggregation and distributed data fusion are proposed to perform the desired aggregation (or fusion) operators en route-eliminating data redundancy, minimizing transmissions and thus saving energy. An operator involved with in-network processing will be placed on a network node, which receives the data from sources, process them and send the output to either the next operator or sink node. As transmitting data from one operator to other imposes a cost, which is dependent on the placement of operators, the placement of operators can greatly affect the energy cost of in-network processing. In this research work, we propose a minimum-cost forwarding based asynchronous algorithm (MCFA) to find the optimal placement for operator tree with minimized energy cost of in-network processing. It is shown that minimum-cost forwarding can dramatically reduce message overhead of asynchronous algorithm. It is also shown that MCFA has less message overhead than synchronous algorithm by both mathematical analysis and simulation-based evaluation. For a regular grid network and a complete binary operator tree, the messages sent at each node are O(√NM) for MCFA, meanwhile O(√NM log2M) for synchronous algorithm, where N is the number of network nodes and M is the number of data objects in operator tree.

In this paper, PAFusion, a power-aware framework for distributed data fusion applications in wireless sensor networks is proposed. It is designed for mapping distributed data fusion application into networks. We have evaluated the performances of PAFusion compared to DFuse and optimal solution through simulation. The simulation results show PAFusion has better performances than DFuse and PAFusion has much less overhead and causes only slightly more transmission cost than optimal solution.

Active Networks paradigm integrated with distributed data fusion has the potential to significantly reduce energy dissipation in WSNs, where energy conservation is the most challenging issue. This work aims to minimize energy cost when distributed data fusion is deployed for Active Networks computing paradigm. An approximate solution with much less computational complexity than optimal solution is proposed for large networks called P2lace, which includes two phases, task graph partition and task graph placement. An extensive experimental evaluation compares approximate solution with optimal solution. The results show that our approximate solution is scalable to different task graph characteristics and network size and only causes little more transmission cost than optimal solution.