The dataset used in this paper is a collection of 30 trajectories generated per gridworld map by randomly selecting for each trajectory the initial locations of both the actor and target.
BibTex:
Before browse our site, please accept our cookies policy
