SSL-powered EFAM: bi-path fusion and channel attention elevate SED accuracy while cutting labeled-data needs

CAS-FEND distills comment knowledge into a content-only student, beating comment-aware rivals with a quarter of comments for timely fake-news catch

Autonomous driving systems increasingly rely on data-driven approaches, yet many still struggle with reasoning, handling rare scenarios, and transparently explaining their actions. A new study introduces DriveMLM, a multi-modal large language model framework that aligns language-based reasoning with structured behavioral planning states, enabling full closed-loop driving in realistic simulators. By integrating multi-view images, LiDAR inputs, traffic rules, and natural-language instructions, DriveMLM generates both driving decisions and human-readable explanations that map directly to vehicle control. The system significantly improves safety, adaptability, and interpretability, demonstrating how large language models (LLMs) can advance the next generation of autonomous driving technology.

Abu Dhabi, United Arab Emirates – The United Arab Emirates has launched Abu Dhabi’s AI Ecosystem for Global Agricultural Development, a platform designed to bring AI solutions to climate-exposed agricultural regions and support the communities most affected by shifting weather patterns.