This pilot study explores the effects of NAO robot-assisted classrooms on autistic students’ classroom performance. Results show NAO significantly improves online attention, communication, interaction assessment, and emotional states. Most students exhibit better engagement and stability, supporting NAO as an effective tool for autism spectrum disorder (ASD) group classroom intervention.

A new study is opening a rare acoustic window into the hidden world of underwater predators. Using machine learning and underwater recordings, researchers developed a system that can detect and classify shell-crushing feeding events by whitespotted eagle rays as they prey on clams, oysters and other shellfish. The technology delivers near-deep learning accuracy with far less computing power, paving the way for real-time, large-scale monitoring of predator behavior and the health of coastal ecosystems.

Scientists repeatedly sampled microbes from six sites up to 1.5 kilometers deep across four years inside a former goldmine. Microbial ecosystems appear to be structured around shared functions rather than shared species. Each ecosystem was organized around two broad groups of microbes: a stable group and a responsive group. Ecosystems were very different among sites but largely stable through time. Study could have implications for underground engineering projects, including carbon storage.

GaN power electronics combines performance, efficiency, and cost advantages and plays a central technological role in the energy transition. At PCIM Expo & Conference 2026 in Nuremberg from June 9 to 11, Fraunhofer IAF will present the latest developments in the field of GaN power electronics—with the highlight being a demonstrator of a bidirectional, single-phase 800 V DC charger for electric vehicles. It was developed by project partner Ambibox as part of the BMWE project GaN4EmoBiL and incorporates a 1200 V GaN module from Fraunhofer IAF.

The ESCUTIC group of the EHU-University of the Basque Country points out that overreliance on tools such as ChatGPT in learning largely depends on each student’s ability to organise him-/herself, work hard and reflect. When key skills such as perseverance, decision-making and learning from mistakes are in place, AI can become a useful aid without replacing independent thinking.

Core-collapse supernovae are thought to be powered by neutrino-driven energy transport. Now, researchers in Japan have shown that neutrino fast flavor conversion can either enhance or suppress supernova explosions depending on the progenitor star. Using advanced multiangle neutrino radiation hydrodynamics simulations, the team demonstrated that the mass accretion rate is the key factor governing this bifurcated behavior, offering new insight into the evolution and death of massive stars.