Neural SLAM: Learning to Explore with External Memory

TL;DR

This paper introduces a neural network model that learns to explore and map new environments by integrating SLAM-like procedures into an external memory architecture, enabling agents to develop long-term spatial memory.

Contribution

It presents a novel differentiable neural architecture that mimics SLAM procedures within an external memory, improving exploration in reinforcement learning agents.

Findings

Effective in grid-world environments

Preliminary success in Gazebo simulations

Facilitates long-term exploration and mapping

Abstract

We present an approach for agents to learn representations of a global map from sensor data, to aid their exploration in new environments. To achieve this, we embed procedures mimicking that of traditional Simultaneous Localization and Mapping (SLAM) into the soft attention based addressing of external memory architectures, in which the external memory acts as an internal representation of the environment. This structure encourages the evolution of SLAM-like behaviors inside a completely differentiable deep neural network. We show that this approach can help reinforcement learning agents to successfully explore new environments where long-term memory is essential. We validate our approach in both challenging grid-world environments and preliminary Gazebo experiments. A video of our experiments can be found at: https://goo.gl/G2Vu5y.