Sim-to-Real Transfer for Vision-and-Language Navigation

TL;DR

This paper explores transferring a vision-and-language navigation agent from simulation to real robots, proposing methods to bridge the sim-to-real gap and evaluating success in unseen environments.

Contribution

It introduces a subgoal model and domain randomization techniques for effective sim-to-real transfer in VLN tasks in robotics.

Findings

Sim-to-real transfer achieves 46.8% success with prior mapping.

Without prior mapping, success drops to 22.5%.

Annotated real environment enables better sim-to-real performance.

Abstract

We study the challenging problem of releasing a robot in a previously unseen environment, and having it follow unconstrained natural language navigation instructions. Recent work on the task of Vision-and-Language Navigation (VLN) has achieved significant progress in simulation. To assess the implications of this work for robotics, we transfer a VLN agent trained in simulation to a physical robot. To bridge the gap between the high-level discrete action space learned by the VLN agent, and the robot's low-level continuous action space, we propose a subgoal model to identify nearby waypoints, and use domain randomization to mitigate visual domain differences. For accurate sim and real comparisons in parallel environments, we annotate a 325m2 office space with 1.3km of navigation instructions, and create a digitized replica in simulation. We find that sim-to-real transfer to an environment not seen in training is successful if an occupancy map and navigation graph can be collected and annotated in advance (success rate of 46.8% vs. 55.9% in sim), but much more challenging in the hardest setting with no prior mapping at all (success rate of 22.5%).