As Multimodal Large Language Models (MLLMs) advance, multimodal agents show promise in real-world tasks like web navigation and embodied intelligence. However, due to limitations in a lack of external feedback, these agents struggle with self-correction and generalization. A promising approach is to use reward models as external feedback, but there is no clear on how to select reward models for agents. Thus, there is an urgent need to build a reward bench targeted at agents. To address these challenges, we propose Agent-RewardBench, a benchmark designed to evaluate reward modeling ability in MLLMs. The benchmark is characterized by three key features: (1) Multiple dimensions and real-world agent scenarios evaluation. It covers perception, planning, and safety with 7 scenarios; (2) Step-level reward evaluation. It allows for the assessment of agent capabilities at the individual steps of a task, providing a more granular view of performance during the planning process; and (3) Appropriately difficulty and high-quality. We carefully sample from 10 diverse models, difficulty control to maintain task challenges, and manual verification to ensure the integrity of the data. Experiments demonstrate that even state-of-the-art multimodal models show limited performance, highlighting the need for specialized training in agent reward modeling. Code is available at github.

Perpetual voting considers sequences of decis ions made by the same electorate, where fairness must be evaluated over time rather than perdecision [16]. A centralchallenge in this setting is ensuring adequaterepresentation for voters who are repeatedly in the minority. Traditional a ggregation rules, such as majority voting or Borda count, fail in this regard: they offer no guarantees of long-term fai rness or cumulative influence. In response, methods such as Perpetual Phragmén [17] and Perpetual Consensus [16] hav e been proposed to distribute influence more equitably over time. However, they rely on full knowledge of all voters ' approval sets, implicitly requiring consistent voter participation, a condition which can be hard to satisfy in real-world contexts. Real-world elections face various practical constraints-- including scheduling conflicts, limited resources, and restricted information access--that inevitably prevent vote rs from participating consistently.

This paper introduces a new approach for approximating the learning dynamics of multiple reinforcement learning (RL) agents interacting in a finite-state Markov game. The idea is to rescale the learning process by simultaneously reducing the learning rate and increasing the update frequency, effectively treating the agent's parameters as a slow-evolving variable influenced by the fast-mixing game state. Under mild assumptions-ergodicity of the state process and continuity of the updates-we prove the convergence of this rescaled process to an ordinary differential equation (ODE). This ODE provides a tractable, deterministic approximation of the agent's learning dynamics. An implementation of the framework is available at\,: https://github.com/yannKerzreho/MarkovGameApproximation

Recent advances in large language models (LLMs) have sparked growing interest in building fully autonomous agents. However, fully autonomous LLM-based agents still face significant challenges, including limited reliability due to hallucinations, difficulty in handling complex tasks, and substantial safety and ethical risks, all of which limit their feasibility and trustworthiness in real-world applications. To overcome these limitations, LLM-based human-agent systems (LLM-HAS) incorporate human-provided information, feedback, or control into the agent system to enhance system performance, reliability and safety. These human-agent collaboration systems enable humans and LLM-based agents to collaborate effectively by leveraging their complementary strengths. This paper provides the first comprehensive and structured survey of LLM-HAS. It clarifies fundamental concepts, systematically presents core components shaping these systems, including environment & profiling, human feedback, interaction types, orchestration and communication, explores emerging applications, and discusses unique challenges and opportunities arising from human-AI collaboration. By consolidating current knowledge and offering a structured overview, we aim to foster further research and innovation in this rapidly evolving interdisciplinary field. Paper lists and resources are available at https://github.com/HenryPengZou/Awesome-Human-Agent-Collaboration-Interaction-Systems.

This technical report presents submission systems for Task 4 of the DCASE 2025 Challenge. This model incorporates additional audio features (spectral roll-off and chroma features) into the embedding feature extracted from the mel-spectral feature to im-prove the classification capabilities of an audio-tagging model in the spatial semantic segmentation of sound scenes (S5) system. This approach is motivated by the fact that mixed audio often contains subtle cues that are difficult to capture with mel-spectrograms alone. Thus, these additional features offer alterna-tive perspectives for the model. Second, an agent-based label correction system is applied to the outputs processed by the S5 system. This system reduces false positives, improving the final class-aware signal-to-distortion ratio improvement (CA-SDRi) metric. Finally, we refine the training dataset to enhance the classi-fication accuracy of low-performing classes by removing irrele-vant samples and incorporating external data. That is, audio mix-tures are generated from a limited number of data points; thus, even a small number of out-of-class data points could degrade model performance. The experiments demonstrate that the submit-ted systems employing these approaches relatively improve CA-SDRi by up to 14.7% compared to the baseline of DCASE 2025 Challenge Task 4.

Pathology is experiencing rapid digital transformation driven by whole-slide imaging and artificial intelligence (AI). While deep learning-based computational pathology has achieved notable success, traditional models primarily focus on image analysis without integrating natural language instruction or rich, text-based context. Current multimodal large language models (MLLMs) in computational pathology face limitations, including insufficient training data, inadequate support and evaluation for multi-image understanding, and a lack of autonomous, diagnostic reasoning capabilities. To address these limitations, we introduce PathChat+, a new MLLM specifically designed for human pathology, trained on over 1 million diverse, pathology-specific instruction samples and nearly 5.5 million question answer turns. Extensive evaluations across diverse pathology benchmarks demonstrated that PathChat+ substantially outperforms the prior PathChat copilot, as well as both state-of-the-art (SOTA) general-purpose and other pathology-specific models. Furthermore, we present SlideSeek, a reasoning-enabled multi-agent AI system leveraging PathChat+ to autonomously evaluate gigapixel whole-slide images (WSIs) through iterative, hierarchical diagnostic reasoning, reaching high accuracy on DDxBench, a challenging open-ended differential diagnosis benchmark, while also capable of generating visually grounded, humanly-interpretable summary reports.

A cooperative circumnavigation framework is proposed for multi-quadrotor systems to enclose and track a moving target without reliance on external localization systems. The distinct relationships between quadrotor-quadrotor and quadrotor-target interactions are evaluated using a heterogeneous perception strategy and corresponding state estimation algorithms. A modified Kalman filter is developed to fuse visual-inertial odometry with range measurements to enhance the accuracy of inter-quadrotor relative localization. An event-triggered distributed Kalman filter is designed to achieve robust target state estimation under visual occlusion by incorporating neighbor measurements and estimated inter-quadrotor relative positions. Using the estimation results, a cooperative circumnavigation controller is constructed, leveraging an oscillator-based autonomous formation flight strategy. We conduct extensive indoor and outdoor experiments to validate the efficiency of the proposed circumnavigation framework in occluded environments. Furthermore, a quadrotor failure experiment highlights the inherent fault tolerance property of the proposed framework, underscoring its potential for deployment in search-and-rescue operations.

We consider an online variant of the fuel-constrained UAV routing problem with a ground-based mobile refueling station (FCURP-MRS), where targets incur unknown fuel costs. We develop a two-phase solution: an offline heuristic-based planner computes initial UAV and UGV paths, and a novel online planning algorithm that dynamically adjusts rendezvous points based on real-time fuel consumption during target processing. Preliminary Gazebo simulations demonstrate the feasibility of our approach in maintaining UAV-UGV path validity, ensuring mission completion. Link to video: https://youtu.be/EmpVj-fjqNY

The proliferation of LLM-based agents has led to increasing deployment of inter-agent collaboration for tasks like scheduling, negotiation, resource allocation etc. In such systems, privacy is critical, as agents often access proprietary tools and domain-specific databases requiring strict confidentiality. This paper examines whether LLM-based agents demonstrate an understanding of contextual privacy. And, if instructed, do these systems preserve inference time user privacy in non-adversarial multi-turn conversation. Existing benchmarks to evaluate contextual privacy in LLM-agents primarily assess single-turn, low-complexity tasks where private information can be easily excluded. We first present a benchmark - MAGPIE comprising 158 real-life high-stakes scenarios across 15 domains. These scenarios are designed such that complete exclusion of private data impedes task completion yet unrestricted information sharing could lead to substantial losses. We then evaluate the current state-of-the-art LLMs on (a) their understanding of contextually private data and (b) their ability to collaborate without violating user privacy. Empirical experiments demonstrate that current models, including GPT-4o and Claude-2.7-Sonnet, lack robust understanding of contextual privacy, misclassifying private data as shareable 25.2\% and 43.6\% of the time. In multi-turn conversations, these models disclose private information in 59.9\% and 50.5\% of cases even under explicit privacy instructions. Furthermore, multi-agent systems fail to complete tasks in 71\% of scenarios. These results underscore that current models are not aligned towards both contextual privacy preservation and collaborative task-solving.

In Social Robot Navigation, autonomous agents need to resolve many sequential interactions with other agents. State-of-the art planners can efficiently resolve the next, imminent interaction cooperatively and do not focus on longer planning horizons. This makes it hard to maneuver scenarios where the agent needs to select a good strategy to find gaps or channels in the crowd. We propose to decompose trajectory planning into two separate steps: Conflict avoidance for finding good, macroscopic trajectories, and cooperative collision avoidance (CCA) for resolving the next interaction optimally. We propose the Probabilistic Gap Planner (PGP) as a conflict avoidance planner. PGP modifies an established probabilistic collision risk model to include a general assumption of cooperativity. PGP biases the short-term CCA planner to head towards gaps in the crowd. In extensive simulations with crowds of varying density, we show that using PGP in addition to state-of-the-art CCA planners improves the agents' performance: On average, agents keep more space to others, create less tension, and cause fewer collisions. This typically comes at the expense of slightly longer paths. PGP runs in real-time on WaPOCHI mobile robot by Honda R&D.