SPARC: Spatial-Aware Path Planning via Attentive Robot Communication
Researchers have developed SPARC, a novel communication mechanism for multi-robot path planning that prioritizes messages from spatially relevant neighbors. This approach, called Relation enhanced Multi Head Attention (RMHA), embeds pairwise Manhattan distances into attention weight computations to improve coordination in dense environments. SPARC was demonstrated to achieve a 75 percent success rate with 128 robots in simulations, significantly outperforming existing methods. AI
IMPACT Enhances coordination in multi-robot systems, potentially improving efficiency in logistics and autonomous operations.