Mapping Navigation Instructions to Continuous Control Actions with Position-Visitation Prediction

Abstract

We propose an approach for mapping natural language instructions and raw observations to continuous control of a quadcopter drone. Our model predicts interpretable position-visitation distributions indicating where the agent should go during execution and where it should stop, and uses the predicted distributions to select the actions to execute. This two-step model decomposition allows for simple and efficient training using a combination of supervised learning and imitation learning. We evaluate our approach with a realistic drone simulator, and demonstrate absolute task-completion accuracy improvements of 16.85% over two state-of-the-art instruction-following methods.

Cite

Text

Blukis et al. "Mapping Navigation Instructions to Continuous Control Actions with Position-Visitation Prediction." Conference on Robot Learning, 2018.

Markdown

[Blukis et al. "Mapping Navigation Instructions to Continuous Control Actions with Position-Visitation Prediction." Conference on Robot Learning, 2018.](https://mlanthology.org/corl/2018/blukis2018corl-mapping/)

BibTeX

@inproceedings{blukis2018corl-mapping,
  title     = {{Mapping Navigation Instructions to Continuous Control Actions with Position-Visitation Prediction}},
  author    = {Blukis, Valts and Misra, Dipendra Kumar and Knepper, Ross A. and Artzi, Yoav},
  booktitle = {Conference on Robot Learning},
  year      = {2018},
  pages     = {505-518},
  url       = {https://mlanthology.org/corl/2018/blukis2018corl-mapping/}
}