ZSON: Zero-Shot Object-Goal Navigation using Multimodal Goal Embeddings

TL;DR

This paper introduces ZSON, a zero-shot object-goal navigation method that leverages multimodal goal embeddings trained on image-goal tasks, enabling agents to find objects in open-world environments using natural language instructions without prior rewards or demonstrations.

Contribution

The paper proposes a novel zero-shot approach for object-goal navigation using multimodal semantic embeddings trained on image-goal data, allowing natural language goal specification in open environments.

Findings

Achieves 4.2% to 20.0% success improvement over existing zero-shot methods.

Enables agents to follow complex and compound natural language instructions.

Generalizes well across multiple datasets and real-world scenarios.

Abstract

We present a scalable approach for learning open-world object-goal navigation (ObjectNav) -- the task of asking a virtual robot (agent) to find any instance of an object in an unexplored environment (e.g., "find a sink"). Our approach is entirely zero-shot -- i.e., it does not require ObjectNav rewards or demonstrations of any kind. Instead, we train on the image-goal navigation (ImageNav) task, in which agents find the location where a picture (i.e., goal image) was captured. Specifically, we encode goal images into a multimodal, semantic embedding space to enable training semantic-goal navigation (SemanticNav) agents at scale in unannotated 3D environments (e.g., HM3D). After training, SemanticNav agents can be instructed to find objects described in free-form natural language (e.g., "sink", "bathroom sink", etc.) by projecting language goals into the same multimodal, semantic embedding space. As a result, our approach enables open-world ObjectNav. We extensively evaluate our agents on three ObjectNav datasets (Gibson, HM3D, and MP3D) and observe absolute improvements in success of 4.2% - 20.0% over existing zero-shot methods. For reference, these gains are similar or better than the 5% improvement in success between the Habitat 2020 and 2021 ObjectNav challenge winners. In an open-world setting, we discover that our agents can generalize to compound instructions with a room explicitly mentioned (e.g., "Find a kitchen sink") and when the target room can be inferred (e.g., "Find a sink and a stove").