We present the Frontier Aware Search with backTracking (FAST) Navigator, a general framework for action decoding, that achieves state-of-the-art results on the Room-to-Room (R2R) Vision-and-Language navigation challenge of Anderson et. al. (2018). Given a natural language instruction and photo-realistic image views of a previously unseen environment, the agent was tasked with navigating from source to target location as quickly as possible. While all current approaches make local action decisions or score entire trajectories using beam search, ours balances local and global signals when exploring an unobserved environment. Importantly, this lets us act greedily but use global signals to backtrack when necessary. Applying FAST framework to existing state-of-the-art models achieved a 17% relative gain, an absolute 6% gain on Success rate weighted by Path Length (SPL).
@article{arxiv.1903.02547,
title = {Tactical Rewind: Self-Correction via Backtracking in Vision-and-Language Navigation},
author = {Liyiming Ke and Xiujun Li and Yonatan Bisk and Ari Holtzman and Zhe Gan and Jingjing Liu and Jianfeng Gao and Yejin Choi and Siddhartha Srinivasa},
journal= {arXiv preprint arXiv:1903.02547},
year = {2019}
}
Comments
CVPR 2019 Oral, video demo: https://youtu.be/AD9TNohXoPA