Learning Composable Behavior Embeddings for Long-horizon Visual Navigation

TL;DR

This paper introduces Composable Behavior Embedding (CBE), a continuous representation for long-horizon visual navigation that improves memory efficiency and robustness in complex environments.

Contribution

It proposes a novel end-to-end learned continuous behavior embedding for long-horizon navigation, capturing path geometry and handling unseen obstacles.

Findings

CBE enables memory-efficient path following.

CBE improves topological mapping robustness.

CBE saves over an order of magnitude in memory usage.

Abstract

Learning high-level navigation behaviors has important implications: it enables robots to build compact visual memory for repeating demonstrations and to build sparse topological maps for planning in novel environments. Existing approaches only learn discrete, short-horizon behaviors. These standalone behaviors usually assume a discrete action space with simple robot dynamics, thus they cannot capture the intricacy and complexity of real-world trajectories. To this end, we propose Composable Behavior Embedding (CBE), a continuous behavior representation for long-horizon visual navigation. CBE is learned in an end-to-end fashion; it effectively captures path geometry and is robust to unseen obstacles. We show that CBE can be used to performing memory-efficient path following and topological mapping, saving more than an order of magnitude of memory than behavior-less approaches.