Link Search Menu Expand Document

Auxiliary Tasks Speed Up Learning PointGoal Navigation

Paper PDF Code

Authors

Joel Ye (Georgia Institute of Technology)*; Dhruv Batra (Georgia Tech & Facebook AI Research); Erik Wijmans (Georgia Tech); Abhishek Das (Facebook AI Research)

Interactive Session

2020-11-18, 12:30 - 13:00 PST | PheedLoop Session

Abstract

PointGoal Navigation is an embodied task that requires agents to navigate to a specified point in an unseen environment. Wijmans et al. showed that this task is solvable in simulation but their method is computationally prohibitive – requiring 2.5 billion frames of experience and 180 GPU-days. We develop a method to significantly improve sample efficiency in learning PointNav using self-supervised auxiliary tasks (e.g. predicting the action taken between two egocentric observations, predicting the distance between two observations from a trajectory, etc.). We find that naively combining multiple auxiliary tasks improves sample efficiency, but only provides marginal gains beyond a point. To overcome this, we use attention to combine representations from individual auxiliary tasks. Our best agent is 5.5x faster to match the performance of the previous state-of-the-art, DD-PPO, at 40M frames, and improves on DD-PPO’s performance at 40M frames by 0.16 SPL. Our code is publicly available at github.com/joel99/habitat-pointnav-aux.

Video

Reviews and Rebuttal

Reviews & Rebuttal


Conference on Robot Learning 2020