Environment Optimization for Multi-Agent Navigation

The dataset used in the paper is a 2-D environment with a starting region S, a destination region D, and an obstacle region ∆. The environment is described by a starting region S, a destination region D, and an obstacle region ∆ without overlap.