Learning to Navigate in Complex Environments

TL;DR

This paper presents a reinforcement learning approach for complex environment navigation, utilizing auxiliary tasks with multimodal inputs to improve data efficiency and performance, achieving near-human results in 3D maze navigation.

Contribution

It introduces a multi-task learning framework combining goal-driven RL with auxiliary depth and loop closure tasks for improved navigation in complex environments.

Findings

Approaches human-level performance in 3D maze navigation.

Auxiliary tasks significantly enhance data efficiency.

Agent implicitly learns key navigation skills.

Abstract

Learning to navigate in complex environments with dynamic elements is an important milestone in developing AI agents. In this work we formulate the navigation question as a reinforcement learning problem and show that data efficiency and task performance can be dramatically improved by relying on additional auxiliary tasks leveraging multimodal sensory inputs. In particular we consider jointly learning the goal-driven reinforcement learning problem with auxiliary depth prediction and loop closure classification tasks. This approach can learn to navigate from raw sensory input in complicated 3D mazes, approaching human-level performance even under conditions where the goal location changes frequently. We provide detailed analysis of the agent behaviour, its ability to localise, and its network activity dynamics, showing that the agent implicitly learns key navigation abilities.