A scientific team consisting of roboticists from Delft University of Technology and biologists from Wageningen University (both The Netherlands) and Carl von Ossietzky University of Oldenburg (Germany) present: “Bee-Nav”, a robot navigation strategy inspired by honeybees. It allows even very small robots to travel far away from home and return successfully, using a neural memory of only 42 kilobytes. In a new environment, the robot first performs a short learning flight near home, just as honeybees do. After that, it can travel away for hundreds of meters and still find its way back. Bee-Nav enables lightweight, safe robots to navigate on their own, opening the door to applications such as butterfly-like drones monitoring greenhouses. The research also offers new insight into how flying insects may find their way home.

The human eye is not merely an optical window to the world, but also a "micro-display" of systemic microcirculation and neural activity. In the current era of big data, the rapid expansion of multi-source and multimodal ophthalmic datasets presents unprecedented opportunities. A critical scientific question emerges: how can Artificial Intelligence (AI) unlock the hidden potential embedded within these vast and heterogeneous datasets?

Recently, a comprehensive review titled "Data-driven computational methods in ophthalmology: A multimodal perspective" was published in the international journal Eye Discovery. From a "data-centric" perspective, the article systematically evaluates the scientific value of multimodal ophthalmic data and analyzes cutting-edge advances and future challenges in AI-driven ophthalmic research.

Global efforts to combat climate change often focus on what we can see: forests, renewable energy, and surface-level farming practices. A new analysis argues that one of the most significant battlegrounds in carbon sequestration lies hidden deep underground. A comprehensive review led by Professor Nanthi Bolan at The University of Western Australia consolidates global knowledge on deep soil carbon, the massive carbon reserve stored more than 30 centimeters below the surface, which has long been excluded from standard carbon accounting.

The paper provides a critical assessment of this enormous carbon sink, noting that deep soil layers (below 30 cm) contain over 850 petagrams of carbon worldwide. This accounts for approximately 50-60% of the total carbon stored in the top meter of soil. While surface soil carbon is closely tied to vegetation and is quick to respond to management changes, deep carbon dynamics are governed by different, more permanent factors.

The Southwest Research Institute-led Ultraviolet Spectrographs (UVS) instruments aboard ESA’s Jupiter Icy Moons Explorer (Juice) spacecraft and NASA’s Europa Clipper made unique observations of interstellar comet 3I/ATLAS in late 2025. SwRI leads the UVS instruments on both spacecraft, simultaneously imaging both hemispheres of the comet and detecting the comet’s ultraviolet emissions.

Microvascular research is providing important biological insights that may help guide how we diagnose and treat patients living with complex chronic disorders such as Long COVID and ME/CFS. The Complex Disorder Alliance (CODA) will support experts to focus on key hypotheses to more quickly determine which patients can benefit from treatment.

Harnessing solar energy to produce usable power is not new, but the technology is constantly evolving and improving. A major development in recent times is the use of perovskite solar cells (PSCs), which are low-weight, highly efficient, flexible solar cells using perovskite (typically a metal-halide material with a specialized structure) crystal structures to absorb light. Though a promising concept, improvements are necessary for PSCs to be able to reach their full potential. Researchers approach these improvements by introducing an additive, 1H-indole-3-carbohydrazide (1H-CBH) to effectively alleviate the main obstacle of PSCs, which are defects leading to loss of energy and efficiency in the cell.

Brain-computer interfaces are moving beyond simple device control toward more dexterous robotic assistance. A new review highlights how noninvasive neural decoding, deep learning, and shared autonomy are helping robotic devices better interpret user intent, reduce control burden, and support future assistive technologies for people with motor impairments and the general population.