On a commercial thoroughfare in old town Pasadena, California, a stone's throw from NASA's Jet Propulsion Laboratory (JPL), you'll find the Neon Retro Arcade. Among its collection of vintage video games is the 1979 Atari classic Asteroids, in which a pixelated spaceship shoots down a barrage of space rocks to stave off fatal collisions. After long days of work at JPL, Amy Mainzer used to rack up high scores on that console. "It was a hoot," she says. It was also apt, considering she oversees a space mission designed to spot dangerous asteroids before they crash into Earth. That mission, the Near-Earth Object (NEO) Surveyor, was conceived in the early 2000s and finally got the green light in 2022. Its components are now being built, tested, and assembled in clean rooms across the United States ahead of its planned launch in September 2027. "We're in the thick of building everything," says Mainzer, NEO Surveyor's principal investigator and now an astronomer at the University of California, Los Angeles (UCLA).

Automatic 3D reconstruction of indoor spaces from 2D floor plans necessitates high-precision semantic segmentation of structural elements, particularly walls. However, existing methods often struggle with detecting thin structures and maintaining geometric precision. This study introduces MitUNet, a hybrid neural network combining a Mix-Transformer encoder and a U-Net decoder enhanced with spatial and channel attention blocks. Our approach, optimized with the Tversky loss function, achieves a balance between precision and recall, ensuring accurate boundary recovery. Experiments on the CubiCasa5k dataset and a proprietary regional dataset demonstrate MitUNet's superiority in generating structurally correct masks with high boundary accuracy, outperforming standard models. This tool provides a robust foundation for automated 3D reconstruction pipelines. To ensure reproducibility and facilitate future research, the source code and the proprietary regional dataset are publicly available at https://github.com/aliasstudio/mitunet and https://doi.org/10.5281/zenodo.17871079 respectively.

In this paper, we present a novel series of Russian information retrieval datasets constructed from the "Did you know..." section of Russian Wikipedia. Our datasets support a range of retrieval tasks, including fact-checking, retrieval-augmented generation, and full-document retrieval, by leveraging interesting facts and their referenced Wikipedia articles annotated at the sentence level with graded relevance. We describe the methodology for dataset creation that enables the expansion of existing Russian Information Retrieval (IR) resources. Through extensive experiments, we extend the RusBEIR research by comparing lexical retrieval models, such as BM25, with state-of-the-art neural architectures fine-tuned for Russian, as well as multilingual models. Results of our experiments show that lexical methods tend to outperform neural models on full-document retrieval, while neural approaches better capture lexical semantics in shorter texts, such as in fact-checking or fine-grained retrieval. Using our newly created datasets, we also analyze the impact of document length on retrieval performance and demonstrate that combining retrieval with neural reranking consistently improves results. Our contribution expands the resources available for Russian information retrieval research and highlights the importance of accurate evaluation of retrieval models to achieve optimal performance. All datasets are publicly available at HuggingFace. To facilitate reproducibility and future research, we also release the full implementation on GitHub.

Thinking LLMs solve complex tasks at the expense of increased compute and overthinking on simpler problems, while non-thinking LLMs are faster and cheaper but underthink on harder reasoning problems. This has led to the development of separate thinking and non-thinking LLM variants, leaving the onus of selecting the optimal model for each query on the end user. We introduce OptimalThinkingBench, a unified benchmark that jointly evaluates overthinking and underthinking in LLMs and also encourages the development of optimally-thinking models that balance performance and efficiency. Our benchmark comprises two sub-benchmarks: OverthinkingBench, featuring simple math and general queries in 72 domains, and UnderthinkingBench, containing 11 challenging reasoning tasks along with harder math problems. Using novel thinking-adjusted accuracy metrics, we extensively evaluate 33 different thinking and non-thinking models and show that no model is able to optimally think on our benchmark. Thinking models often overthink for hundreds of tokens on the simplest user queries without improving performance. In contrast, large non-thinking models underthink, often falling short of much smaller thinking models. We further explore several methods to encourage optimal thinking, but find that these approaches often improve on one sub-benchmark at the expense of the other, highlighting the need for better unified and optimal models in the future.

Hyabusa2 is already 105.5 million miles on its journey, but new data indicates trouble. Breakthroughs, discoveries, and DIY tips sent every weekday. Japan's Hyabusa2 space probe is currently about 105.5 million miles away, en route to its second asteroid rendezvous . However, revised data collected from a global network of observatories now indicates that the space rock designated as 1998 KY26 will look and behave far differently than astronomers previously theorized--and it may prove disastrous for the tiny explorer. In 2010, the Japanese Aerospace Exploration Agency (JAXA) made history when its Hayabusa probe became the first spacecraft to not only land on and launch from an asteroid (Itokawa), but successfully return to Earth with samples .

With the rapid advancement of multimodal retrieval and its application in LLMs and multimodal LLMs, increasingly complex retrieval tasks have emerged. Existing methods predominantly rely on task-specific fine-tuning of vision-language models and are limited to single-modality queries or documents. In this paper, we propose ReT-2, a unified retrieval model that supports multimodal queries, composed of both images and text, and searches across multimodal document collections where text and images coexist. ReT-2 leverages multi-layer representations and a recurrent Transformer architecture with LSTM-inspired gating mechanisms to dynamically integrate information across layers and modalities, capturing fine-grained visual and textual details. We evaluate ReT-2 on the challenging M2KR and M-BEIR benchmarks across different retrieval configurations. Results demonstrate that ReT-2 consistently achieves state-of-the-art performance across diverse settings, while offering faster inference and reduced memory usage compared to prior approaches. When integrated into retrieval-augmented generation pipelines, ReT-2 also improves downstream performance on Encyclopedic-VQA and InfoSeek datasets. Our source code and trained models are publicly available at: https://github.com/aimagelab/ReT-2

The paper proposes an efficient algebraic solution to the problem of forward kinematics for a general Stewart-Gough platform. The problem involves determining all possible postures of a mobile platform connected to a fixed base by six legs, given the leg lengths and the internal geometries of the platform and base. The problem is known to have 40 solutions (whether real or complex). The proposed algorithm consists of three main steps: (i) a specific sparse matrix of size 293x362 (the elimination template) is constructed from the coefficients of the polynomial system describing the platform's kinematics; (ii) the PLU decomposition of this matrix is used to construct a pair of 69x69 matrices; (iii) all 40 solutions (including complex ones) are obtained by computing the generalized eigenvectors of this matrix pair. The proposed algorithm is numerically robust, computationally efficient, and straightforward to implement - requiring only standard linear algebra decompositions. MATLAB, Julia, and Python implementations of the algorithm will be made publicly available.

Topological localization is a fundamental problem in mobile robotics, since robots must be able to determine their position in order to accomplish tasks. Visual localization and place recognition are challenging due to perceptual ambiguity, sensor noise, and illumination variations. This work addresses topological localization in an office environment using only images acquired with a perspective color camera mounted on a robot platform, without relying on temporal continuity of image sequences. We evaluate state-of-the-art visual descriptors, including Color Histograms, SIFT, ASIFT, RGB-SIFT, and Bag-of-Visual-Words approaches inspired by text retrieval. Our contributions include a systematic, quantitative comparison of these features, distance measures, and classifiers. Performance was analyzed using standard evaluation metrics and visualizations, extending previous experiments. Results demonstrate the advantages of proper configurations of appearance descriptors, similarity measures, and classifiers. The quality of these configurations was further validated in the Robot Vision task of the ImageCLEF evaluation campaign, where the system identified the most likely location of novel image sequences. Future work will explore hierarchical models, ranking methods, and feature combinations to build more robust localization systems, reducing training and runtime while avoiding the curse of dimensionality. Ultimately, this aims toward integrated, real-time localization across varied illumination and longer routes.

Ukraine's European allies marshalled resources this week to provide the besieged country with air defences against drones and ballistic missiles. The European Union also announced an 18th round of sanctions designed to sever all remaining Russian energy imports, and proposed a fivefold increase in the common defence budget to boost EU defence research and procurement. European leaders convinced the United States to symbolically rejoin the 52-nation Ukraine Defence Contact Group coordinating defence donations, but not as a donor. It was the first such meeting attended by US Defense Secretary Pete Hegseth since February, when he told EU members that pushing Russia out of Ukraine's internationally recognised territory was unrealistic. As the ideological chasm between the EU and the US over Ukraine widened, Russia continued to pound Ukrainian defenders, making a few inroads.

Multiperspective Fusion (MPF) is a novel posttraining alignment framework for large language models (LLMs) developed in response to the growing need for easy bias mitigation. Built on top of the SAGED pipeline, an automated system for constructing bias benchmarks and extracting interpretable baseline distributions, MPF leverages multiperspective generations to expose and align biases in LLM outputs with nuanced, humanlike baselines. By decomposing baseline, such as sentiment distributions from HR professionals, into interpretable perspective components, MPF guides generation through sampling and balancing of responses, weighted by the probabilities obtained in the decomposition. Empirically, we demonstrate its ability to align LLM sentiment distributions with both counterfactual baselines (absolute equality) and the HR baseline (biased for Top Univeristy), resulting in small KL divergence, reduction of calibration error and generalization to unseen questions. This shows that MPF offers a scalable and interpretable method for alignment and bias mitigation, compatible with deployed LLMs and requiring no extensive prompt engineering or finetuning.