A new study from Tel Aviv University has predicted, for the first time, the groundbreaking results that can be obtained from detecting radio waves coming to us from the early Universe. The findings show that during the cosmic dark ages, dark matter formed dense clumps throughout the Universe, which pulled in hydrogen gas and caused it to emit intense radio waves. This leads to a novel method to use the measured radio signals to help resolve the mystery of dark matter.

A recent study published in National Science Review conducted change detection in the youngest, topographically steepest, and theoretically most unstable regions on the lunar surface, revealing a large number of new landslides formed since 2009. Endogenic moonquakes rather than new impacts are the primary trigger, and the Imbrium basin may host an active seismic zone.

During the preliminary design phase of flapping-wing micro air vehicles (FWMAVs), there currently exists a deficiency in rapid prediction method for the aerodynamic characteristics of flexible flapping wings. A novel aerodynamic prediction method for flexible flapping wings has recently achieved significant breakthroughs. This method innovatively employs conical surface to mimic wing deformation, combined with an unsteady panel method for aerodynamic force computation, enabling rapid and accurate prediction of both aerodynamic characteristics and control moments of flexible flapping wings.

Unmanned Swarm Systems (USS) have transformed key fields like disaster rescue, transportation, and military operations via distributed coordination, yet trajectory prediction accuracy and interaction mechanism interpretability remain major bottlenecks—issues that existing methods fail to address by either ignoring physical constraints or lacking explainability. A recent breakthrough from Northwestern Polytechnical University solves this: Dr. Shuheng Yang and Prof. Dong Zhang developed the Swarm Relational Inference (SRI) model, an unsupervised end-to-end framework integrating swarm dynamics with dynamic graph neural networks. This model not only enhances interpretability and physical consistency but also drastically reduces long-term prediction errors, marking a critical step toward reliable autonomous collaboration for real-world USS applications.