Cooperative Vision-and-Dialog Navigation

Cooperative Vision-and-Dialog Navigation (CVDN) is a dataset of embodied, human-human dialogs situated in a simulated, photorealistic home environment. The Navigator asks questions of their partner, the Oracle, who has privileged access to the best next steps the Navigator should take according to a full-state information shortest path planner. The dataset consists of 2050 human-human navigation dialogs, comprising over 7k navigation trajectories punctuated by question-answer exchanges across 83 house scans.
@inproceedings{thomason:corl19,
  title={Vision-and-Dialog Navigation},
  author={Jesse Thomason and Michael Murray and Maya Cakmak and Luke Zettlemoyer},
  booktitle={Conference on Robot Learning (CoRL)},
  year={2019}
}

CVDN Example Dialog