New AI model helps robot teams navigate crowded spaces more effectively

Researchers have developed a new AI model called HMAGAT that helps multiple robots navigate to goals without collisions. It uses hypergraphs to model complex group interactions, which standard AI models miss. Despite being 85 times smaller and trained on 100 times less data, it outperforms the current best model. This shows that designing AI with the right structure for a task is often more important than using massive datasets or models.