Spatial Relation Graph and Graph Convolutional Network for Object Goal Navigation

TL;DR

This paper introduces a novel framework combining a Spatial Relational Graph and Graph Convolutional Networks to improve object-goal navigation by modeling spatial relationships and object likelihoods, enabling robots to efficiently locate target objects.

Contribution

The work presents a new approach that integrates SRG and GCN for spatial reasoning in object navigation, enhancing the robot's ability to estimate and explore target regions.

Findings

Improved navigation accuracy in object-goal tasks

Effective use of SRG and GCN for spatial reasoning

Enhanced target object localization efficiency

Abstract

This paper describes a framework for the object-goal navigation task, which requires a robot to find and move to the closest instance of a target object class from a random starting position. The framework uses a history of robot trajectories to learn a Spatial Relational Graph (SRG) and Graph Convolutional Network (GCN)-based embeddings for the likelihood of proximity of different semantically-labeled regions and the occurrence of different object classes in these regions. To locate a target object instance during evaluation, the robot uses Bayesian inference and the SRG to estimate the visible regions, and uses the learned GCN embeddings to rank visible regions and select the region to explore next.