Vision-Language Navigation Policy Learning and Adaptation

  • Xin Wang ,
  • Qiuyuan Huang ,
  • Asli Celikyilmaz ,
  • ,
  • Dinghan Shen ,
  • Yuan-fang Wang ,
  • William Wang ,
  • Lei Zhang

IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI), 2020 |

Publication

Vision-language navigation (VLN) is the task of navigating an embodied agent to carry out natural language instructions inside real 3D environments. In this paper, we study how to address three critical challenges for this task: the cross-modal grounding, the ill-posed feedback, and the generalization problems. First, we propose a novel Reinforced Cross-Modal Matching (RCM) approach that enforces cross-modal grounding both locally and globally via reinforcement learning (RL). Particularly, a matching critic is used to provide an intrinsic reward to encourage global matching between instructions and trajectories, and a reasoning navigator is employed to perform cross-modal grounding in the local visual scene. Evaluation on a VLN benchmark dataset shows that our RCM model significantly outperforms baseline methods by 10% on Success Rate weighted by Path Length (SPL) and achieves the state-of-the-art performance. To improve the generalizability of the learned policy, we further introduce a Self-Supervised Imitation Learning (SIL) method to explore and adapt to unseen environments by imitating its own past, good decisions. We demonstrate that SIL can approximate a better and more efficient policy, which tremendously minimizes the success rate performance gap between seen and unseen environments (from 30.7% to 11.7%).