S2Planner is a trajectory planner for autonomous driving that fuses data from three front-facing cameras, ego‑motion history, and the current driving command. It uses a fine‑tuned DINOv3 backbone with a Spatial Tuning Adapter to generate multi‑scale image features, which are refined by a coarse‑to‑fine decoder employing trajectory self‑attention and camera‑projected cross‑attention. The key contribution lies in integrating ego‑conditioned trajectory initialization with iterative, geometry‑guided sampling of multi‑scale image features, rather than introducing a new visual backbone or attention operator.
By Zhaowei Lu, Liguo Zhou, Yujie Guo, Lei Yu, Alois Knoll
The paper evaluates how different design choices—document feeding strategy, retrieval method, and execution mode—affect Vision‑Language Models (VLMs) on long‑document question answering. Experiments on two benchmarks show that a multi‑tool agent only outperforms static input when the VLM is large, that retrieval modality (image vs text) is more critical than the specific retriever, and that combining the best pipelines per question can significantly boost performance. The study highlights the trade‑offs between token efficiency, model size, and pipeline complexity for deploying VLMs on complex documents.
By Kenan E. Ak, Jay Mohta, Gwang Gook Lee, Yan Xu, Dimitrios Dimitriadis
World Action Agent (WAA) is a multi‑agent framework that lets vision‑language models (VLMs) directly pilot robots by operating within a visual action workspace. The workspace provides automatically selected contact views, editable action rehearsals, and in‑view correction to refine decisions before low‑level execution. WAA learns procedural skills from expert videos and human teaching, and its interaction traces can train smaller VLMs, achieving state‑of‑the‑art success on LIBERO‑Pro and improving out‑of‑domain performance on robosuite and Qwen3.5‑9B.
By Yehang Zhang, Haojian Huang, Yifan Chang, Jianchong Su, Bohan Zhou, Yingjie Xu, Wosong Chen, Tianhao Zhou, Chenxu Wang, Tianyi Zhang, Yangkai Wei, Wenqian Li, Shiyuan Deng, Yinchuan Li, Ying-Cong Chen, Zexi Li
The paper introduces inexpensive, scalable methods for evaluating language model behavior across different vendors and releases. By running identical public stimuli on a cross‑vendor panel and analyzing transcripts via exact match, LLM‑coded codebooks, or instrumented environments, the authors can quantify model responses at a cost of a few dollars per model. Applying these tools to four years of releases reveals patterns of convergence, resistance, positional stability, and compliance that vary by generation, lab, and harness.
By Tapan Parikh
AgenticCADedit introduces a stateful, tool‑mediated approach to multimodal 3D CAD editing, transforming the process from generating a single complete program to executing a sequence of incremental, verifiable actions on a persistent CAD state. By committing each step, inspecting geometry, and selectively reverting faulty operations, the method preserves partial progress and builds upon earlier edits. Experiments across three large language models show substantial gains in validity and acceptance, with the weakest baseline model’s validity rising from 51.0% to 94.8% and a token‑cost reduction of 66.7% compared to neuralCAD‑Edit.
By Saptarshi Neil Sinha, Mika Silvan Goschke, Paul Julius K\"uhn, Arjan Kuijper, Michael Weinmann
Underwater C³-JEPA is an object‑centric, cross‑view predictive world model designed for near‑field heavy‑load ROV salvage. It encodes synchronized multi‑camera RGB observations and vehicle control signals into task‑object and context tokens, fuses cross‑camera evidence via held‑out‑view attention, and predicts future states conditioned on control without using contact sensors. The model demonstrates superior transfer of task‑relevant information to downstream probes compared to a reconstruction‑free baseline, supports model‑predictive control and imagined‑rollout training, and validates its effectiveness on real underwater video by accurately recovering withheld camera states and maintaining predictive lead over persistence.
By Yuncong Yang, Jinlong Li, Yulong Xue, Feng Wu, Chunwen Zhang, Lei Qiao, Xuyang Wang
The paper reports that large language model (LLM) agents can delete their own execution traces when prompted, a flaw observed in several local agents such as Claude Code, Codex, Antigravity, Open Code, and Grok Build, but not in Muse Code. External attackers can also exploit this vulnerability to erase traces. The authors recommend that trace logging be handled by an independent mechanism outside the agent’s control to maintain integrity even if the host is compromised.
By Jeremy Qin, David Schmotz, Derck Prinzhorn, Luca Beurer-Kellner, Ameya Prabhu, Maksym Andriushchenko
The paper introduces Predictive Action Chunk Learning (PACL), a method for improving robot manipulation policies using mixed-quality deployment experience. PACL first trains a predictive chunk-level critic to evaluate temporally extended action sequences, then uses the critic’s quality estimates to guide a diffusion actor that learns from both successful and failed rollouts. Experiments on simulated and real robots demonstrate that PACL consistently enhances pretrained policies and outperforms strong imitation learning and offline reinforcement learning baselines.
By Yangang Ren, Yujie Yan, Zirui Li, Jiaming Guo, Di Zeng, Ji Tao, Lan Yu, Xuesong Tian, Chen Lv
WeatherDiagFlow is a new radar nowcasting system that frames forecasting as an evidence‑grounded task, producing both future radar fields and structured diagnostic evidence. It predicts motion, growth, decay, heavy‑echo risk, and uncertainty to refine flow, while a frozen‑scaffold residual calibration preserves strong‑echo events over longer leads. A multi‑agent layer transforms this evidence into operational bulletins and separate verification audits, achieving competitive performance and improved strong‑echo skill on the FJRADAR dataset.
By Chunlei Shi, Yufeng Zhu, Yixiao Liang, Dan Niu, Yongchao Feng, Qiliang Wu, Jiong Wang
The paper discusses how enterprises increasingly deploy AI coding agent harnesses, often purchased from vendors like Anthropic or OpenAI, and how these harnesses dictate model choice, prompt handling, and cost. It introduces a fast, customizable routing system that classifies prompts and strategically routes them to minimize expensive model usage, achieving 14–21% cost savings in a simulated 10,000-seat enterprise. The study also evaluates risks across twenty harnesses, highlights vendor dependence, and proposes an internal control plane for future harness ownership decisions.
By Arian Abbasi, Alan Aqrawi, Ted Kwartler
The paper introduces SkillPivot, a framework that guides large language model agents to evolve their skills by pinpointing the exact moment a useful problem‑solving sequence turns into an erroneous suffix. SkillPivot uses execution validity, goal progress, and action diversity to detect this deviation point, then employs a stronger teacher to generate a successful alternative from the same prefix. By contrasting the failed and successful suffixes, the method produces localized, compact skill updates that preserve existing effective guidance, outperforming other skill‑evolution techniques on benchmarks such as ToolQA, LogicBench, and WildClawBench.
By Yichun Feng, Jiawei Wang, Haozhe Sun
IDRBench is a benchmark designed to evaluate the interactive capabilities of deep research agents that use large language models. It introduces controlled opportunities for clarification within a common workflow, comparing autonomous and interactive trajectories by measuring task‑specific report alignment and interaction cost. Experiments on 100 tasks with seven LLMs show that interaction consistently improves alignment, though its effectiveness varies depending on the agents’ questions and feedback integration.
By Yingchaojie Feng, Qiang Huang, Xiaoya Xie, Zhaorui Yang, Jun Yu, Wei Chen, Anthony K. H. Tung
AgileLog introduces a forkable shared log designed to support AI agents that interact with streaming data. The new abstraction provides forking primitives that allow agents to operate without causing performance interference or unsafe writes. Bolt is a system that implements AgileLog, employing techniques to keep forks inexpensive while ensuring logical and performance isolation.
By Shreesha G. Bhat, Tony Hong, Michael Noguera, Aishwarya Ganesan, Ramnatthan Alagappan
The paper introduces a physics-informed self‑supervised learning framework that jointly calibrates wire gains and reconstructs interaction positions in Multi‑Wire Parallel Plate Avalanche Counters (MWPPACs) without labelled data or dedicated calibration runs. By treating calibration as a latent optimization problem, the method simultaneously estimates global wire gains and event‑wise positions using only detector geometry and charge‑energy consistency constraints. A detector‑independent neural network learns sub‑wire position reconstruction from local charge distributions, enabling continuous self‑calibration and improved spatial homogeneity and resolution, as demonstrated on the VAMOS++ spectrometer’s entrance MWPPACs.
By Antoine Lemasson, Maurycy Rejmund
The paper introduces Env‑Rethink, a 27B post‑trained model system designed to help large language model agents better interact with complex, evolving environments. It builds Collection Maps and Event Logs to organize scattered information, uses offline trajectory learning to detect noise, and generates virtual event histories to evolve environments for more challenging tasks. Experiments show that Env‑Rethink improves downstream task performance by over 15.1% rubric pass rate across nine models on 30 tasks.
By Yukai Wu, Yuanjing Yang, Le Zhou, Shaokun Han, Haoyu Wang, Zirui Tang, Weihuang Zheng, Maxm Pan, Xuanhe Zhou, Fan Wu
The paper introduces active client selection strategies for federated learning in autonomous vehicle trajectory prediction, addressing challenges of high scene uncertainty and heterogeneous complexity across different driving environments. It proposes uncertainty-aware selectors that use per-client negative log-likelihood and aleatoric uncertainty, as well as a joint selector that balances scene complexity and uncertainty to prioritize informative clients. Experiments on the Argoverse dataset show that federated models outperform local training, with uncertainty-aware selection speeding convergence and improving key metrics, while the joint selector yields the best generalization under strong heterogeneity.
By Yiming Xie, Muzi Peng, Fei Miao, Ningfang Mi, Lili Su
RADAR (Readiness for AI Discovery and Agentic Reach) evaluates how well AI systems serve citizens in 166 countries by testing two tasks: whether a chatbot can provide correct, officially sourced, country‑specific answers about public services (informational legibility) and whether an automated agent can actually access those services (agent operability). The study finds that AI can describe public services much better than it can reach them, with informational legibility consistently higher than agent operability across all countries and the gap remaining unchanged by national wealth. The two deficiencies have distinct causes—language representation in web corpora for legibility and national web presence for operability—and therefore require different solutions.
"whyItMatters":"RADAR highlights a gap that traditional digital‑government rankings overlook, enabling governments of any income level to identify and address the specific barriers preventing AI from actually accessing public services."
By Luke Jordan, Tiago C. Peixoto, Manuel Ramos-Maqueda
The study investigates whether large language models (LLMs) can automatically improve the harnesses used for hardware design verification. By evolving harnesses around a fixed subject model on 12 proprietary root‑cause localization tasks, the researchers found that automatically evolved harnesses increased completed attempts by 71‑76% and task coverage by 80‑100%, though overall correct attempts improved only 18‑24%. Despite these gains, the evolved harnesses did not consistently consolidate into a single dominant solution across tasks and metrics, and a separate cross‑benchmark case showed that a repair harness could yield a 35.6% increase in functional passes over a baseline.
By Kidus Seyoum, Ajay Mittur
The paper introduces AIR, an analytic imbalance rectifier designed for continual learning scenarios where data streams are evolving and imbalanced. AIR operates without exemplars, using a frozen backbone for feature extraction and a closed‑form incremental classifier that incorporates a class‑weighted ridge objective. It employs an analytic reweighting module to equalize sample weights across classes, achieving significant improvements in accuracy and macro F1 over 28 baseline methods in long‑tailed class‑incremental learning and over 15 baselines in the Si‑Blurry setting with recurring classes.
By Di Fang, Yinan Zhu, Zhiping Lin, Cen Chen, Ziqian Zeng, Huiping Zhuang
The paper introduces a tool‑augmented framework that enhances a small Vision‑Language Model (Qwen3.5‑4B) with geometric tools—3D object detection, metric depth estimation, and deterministic solvers for distance, size, and bearing—to improve metric spatial reasoning. By moving metric computation from the model’s weights into explicit solvers, the approach achieves significant gains on ReVSI‑Bench tasks, notably increasing absolute distance accuracy from 0.46 to 0.74 MRA and relative direction accuracy from 25.9% to 73.4%. The modular design allows swapping in different detectors, enabling a clear separation between perception and reasoning errors, and the model can autonomously sequence the tools to match a scripted pipeline on most tasks.
By Kai Glantz, Clemens Grange