OdorSpaces is a physics‑based simulation platform that extends Habitat‑Sim with 3D odor transport and dispersion modeled by Fire Dynamics Simulator, allowing agents to navigate photorealistic indoor environments using both visual and olfactory data. The benchmark presents two source‑seeking tasks—an invisible gas‑leak and an early‑stage fire—providing a realistic testbed for visual‑olfactory embodied navigation. An initial multimodal reinforcement learning policy that fuses visual observations with temporal olfactory measurements outperforms several baselines, establishing a strong reference for future research.
By Changyeon Lee, Kelin Yu, Zheng Wei, Shuna Ni, Ruohan Gao
The paper introduces Consultation Value Distillation (CVD), a method for estimating when a large language model (LLM) should seek human input given limited consultation opportunities. CVD uses offline Monte Carlo rollouts to learn executor-dependent consultation values, then distills these into lightweight predictors that enable efficient budgeted allocation without further rollouts. Experiments across multiple batch and sequential settings show that CVD outperforms direct prompting and Value of Information approaches, achieving near‑lossless allocation with lower search costs.
By Shuli Jiang, Khushbu Pahwa, Yuting Zhang, Yi Zhang, Wei Xia, Stefano Soatto
CellCanon is an agentic cellular world model that integrates single‑cell expression prediction, evidence‑grounded explanation, and observation‑driven updating. Its predictive core reuses condition‑level biological evidence to generate gene‑specific readouts, while a shared world state links predictions, measurements, and source‑linked claims to support explanation and revision. On 339 held‑out SciPlex3 conditions across three cell lines, CellCanon outperforms baselines in differential‑expression AUPRC, Macro‑F1, and direction Macro‑F1, and its explanation module correctly answers all answerable questions with supporting proofs.
By Zhaokang Liang, Shuyang Zhuang, Xiaoran Jiao, Hao Chen
Confucius4‑R2T2, a stable‑prefix streaming ASR model, is introduced to ensure that once a token is emitted it is never revised, thereby reducing wasted work and incorrect actions in real‑time transcription. The model is derived by post‑training Qwen3‑ASR‑1.7B on 1,500 hours of Chinese and English audio while keeping its audio encoder frozen, and it leverages a new concept called Longest Stable Prefix (LSP) to convert a full‑context model into a stable‑prefix streaming one without architectural changes. Evaluation shows that R2T2 achieves near full‑context quality with low latency and stable output, and its benefits generalize beyond the languages used in training.
By Daniel Huang, Qi He, Shaoxiong Duan, Yingjie Li, Qiao Cheng, Yitao Duan
CoSafe is a modular, ADS-agnostic framework that extends runtime safety assurance from ego-centric takeover to scene-level cooperative intervention in mixed-traffic autonomous driving. It uses vehicle-to-everything communication to allow roadside units to monitor hazards and jointly mitigate risks involving multiple vehicles. Evaluation on the InterDrive benchmark shows that CoSafe improves driving scores by 70.03%, reduces safety violations by 87.80%, and accurately identifies unsafe autonomous vehicles with an F3 score of 0.930 while meeting real-time constraints.
By Dingji Wang, You Lu, Haowen Jiang, Yanggang Chen, Yuyang Bi, Bihuan Chen, Shenglin Bao, Xin Peng
AgentTrustCI is a Java/JUnit 5 framework that turns application‑layer expectations—such as tool selection, grounding of answers, and adherence to workflows—into policy, grounding, and trajectory checks over recorded traces. It provides deterministic replay for repeatable build‑time tests and a judge interface for optional semantic evaluation. In tests across 55 synthetic cases from seven enterprise domains, the full pipeline correctly classified all injected faults and passing controls, while single‑layer configurations handled 35 cases, demonstrating the framework’s coverage and repeatability.
By Sachin Gupta
The paper introduces AutoMMemo, a framework that lets agents automatically design memory mechanisms for learning from multimodal experience. Each memory mechanism is represented as an executable memo program that specifies how past episodes are stored, organized, and retrieved. Through iterative evaluation, reflection-guided mutation, and budget-aware tree search, AutoMMemo learns effective memory designs that improve agent performance on GUI/Web navigation and multimodal visual reasoning tasks while keeping token and interaction costs reasonable.
By Xingyu Sui, Weixiang Zhao, Yongxin Tang, Yanyan Zhao, Yang Wu, Dandan Tu, Bing Qin
The paper introduces the DLP curve, a metric that quantifies how well nonlinear embeddings like t‑SNE and UMAP preserve local and global structure by measuring the share of each point’s k‑nearest neighbors retained in the map. Corrected for chance, the curve yields local fidelity (L) at k=10 and global fidelity (G) over 30–50% of the data, with the midpoint provably equivalent to a rank correlation of distances. The authors also present the LG reversal test to compare two maps and the entity fidelity (E) to predict which entities a map loses, demonstrating strong agreement with established global scores on twelve large datasets.
By Henry Han
The paper "Signals: Continuous Diagnostics of AI Agents" introduces a method for diagnosing problems in AI agents by analyzing agentic traces. It defines 45 problem types across eight families, categorizes findings into engineering issues called Signals, and assesses their severity. The method was evaluated on public agent executions, achieving a 91.43% F1 score in matching Signals to reference issues, outperforming a one-pass LLM control by about 18 percentage points.
By Amey Rambatla, Vatsal Goel, Sagar Gaur, Siddhant Loya, Anurag Mishra, Asif Mohammad
arXiv:2509.10127v3 Announce Type: replace-cross
Abstract: Recent advances in large language models (LLMs) have enabled large-scale, high-fidelity social simulations, creating new opportunities for co...
By Zhengyu Hu, Jianxun Lian, Zheyuan Xiao, Max Xiong, Tianfu Wang, Teng Xiao, Yuxuan Lei, Fengqing Jiang, Kaize Ding, Ziang Xiao, Nicholas Jing Yuan, Xing Xie, Radha Poovendran
arXiv:2510.27428v2 Announce Type: replace-cross
Abstract: Soft robots offer adaptability and safety in unstructured environments, yet their compliant, high-dimensional, and nonlinear dynamics make mo...
By Hehui Zheng, Bhavya Sukhija, Chenhao Li, Klemens Iten, Andreas Krause, Robert K. Katzschmann
arXiv:2610.26826v2 Announce Type: replace-cross
Abstract: Beyond basic tool use and coding skills, effectively incorporating visual references into agentic generation is equally essential for applyin...
By Yuchi Wang, Rongyao Fang, Weihao Yu, Xilin Wei, Tiancheng Gu, Dingkang Yang, Hang Zhang, Dayiheng Liu, Hongsheng Li, Shuai Bai
arXiv:2611.06483v1 Announce Type: new
Abstract: While multi-turn agentic reinforcement learning (RL) excels on long-horizon tasks, its performance gains remain highly sensitive to interaction constra...
By Kejian Zhu, Zhuoran Jin, Shangqing Tu, Tianyi Men, Yupu Hao, Jiachun Li, Kang Liu, Jun Zhao
arXiv:2611.05801v1 Announce Type: cross
Abstract: Holistic perception for autonomous agents requires reliably segmenting and associating scene elements across time and camera views, including objects...
By Rohit Mohan, Maximilian Luz, Swarnava Chowdhury, Yao Lu, Florian Drews, Thomas N\"urnberg, Yakov Miron, Federico Tombari, Stefano Gasperini, Abhinav Valada
arXiv:2611.06854v1 Announce Type: cross
Abstract: Computer-use agents (CUAs) have recently achieved superhuman performance on several long-horizon benchmarks. However, their computational cost limits...
By Pranjal Aggarwal, Sean Welleck
arXiv:2611.06914v1 Announce Type: cross
Abstract: Researchers use large language models as agents in social simulations, where agents are prompted with observations and background information intende...
By Mattia Greiche, Reihaneh Rabbany, Maximilian Puelma Touzel
arXiv:2508.09275v3 Announce Type: replace
Abstract: Collaborative multi-agent reinforcement learning (MARL) is increasingly deployed in critical domains, yet its vulnerability to adversarial perturba...
By Amine Andam, Jamal Bentahar, Mustapha Hedabou
arXiv:2611.02102v3 Announce Type: replace
Abstract: Scaling computer-use agents from individual actions to complex software work requires experience that connects decisions to complete outcomes. We p...
By Chaoyun Zhang, Bingxuan Li, Pinxue Guo, Yangfan Gao, Zikun Cai, Tianci Xie, Hui Li, Yuqi Huang, Yifei Cao, Bo Mao, Tian Tang, Qichang Zheng, Tongtong Liu, Xiaoxing Wang, Jinxiu Liu, Lizhi Lin, Zhou Shao, Maosen Zhou, Miao Zheng, Shuo Zhang, Yangji He, Kepeng Lei, Ning Zhao, Wei Shen, Zhenjiang Jin, Xinyue Zhang, Peng Sun, Zhenhua Han, Rui Zheng, Hang Yan
arXiv:2611.05266v2 Announce Type: replace
Abstract: Naive on-policy RL is often bottlenecked by the capabilities of the base model --- insufficient coverage can result in an inability to learn from c...
By Kevin Frans, Alexander M Rush, Sergey Levine, Pieter Abbeel, Ashvin Nair
arXiv:2611.05731v1 Announce Type: new
Abstract: Most memory systems for LLM agents are retrospective, preserving and retrieving past information to inform current decisions. Proactive agents, however...
By Yuanchen Bei, Zhining Liu, Tianxin Wei, Weizhi Zhang, Zhichen Zeng, Philip S. Yu, Hendrik Hamann, Jingrui He, Hanghang Tong