How to Train PointGoal Navigation Agents on a (Sample and Compute) Budget

By Erik Wijmans, Irfan Essa, Dhruv Batra. 2020

10 citations recorded by the source corpus.

PointGoal navigation has seen significant recent interest and progress, spurred on by the Habitat platform and associated challenge. In this paper, we study PointGoal navigation under both a sample budget (75 million frames) and a compute budget (1 GPU for 1 day). We conduct an extensive set of experiments, cumulatively totaling over 50,000 GPU-hours, that let us identify and discuss a number of ostensibly minor but significant design choices -- the advantage estimation procedure (a key component in training), and visual encoder architecture. Overall, these design choices lead to considerable and consistent improvements. Under a sample budget, performance for RGB-D agents improves 3 SPL on Gibson (4% relative improvement) and 20 SPL on Matterport3D (43% relative improvement). Under a compute budget, performance for RGB-D agents improves by 3 SPL on Gibson (5% relative improvement) and 15 SPL on Matterport3D (50% relative improvement). Our findings and recommendations will serve to make the community's experiments more efficient -- to reach 50 SPL with RGB-D on Matterport3D, they reduce the samples needed by 3x and the training time 2x.

Subfields: Reinforcement Learning, Robotics & Control

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Reinforcement Learning

In Robotics & Control

Citations in the corpus

Habitat 2.0: Training Home Assistants to Rearrange their Habitat (2021, background): “This is consistent with findings in the navigation literature [12] and fortuitous since depth sensors are faster to render than RGB.”