Disentangled Acoustic Fields For Multimodal Physical Scene Understanding

TL;DR

This paper introduces a disentangled acoustic field model for multimodal physical scene understanding, enabling better localization of fallen objects through explicit sound generation and propagation modeling.

Contribution

It proposes a novel disentangled acoustic field model and an analysis-by-synthesis framework for improved sound-based object localization in embodied agents.

Findings

Disentangled acoustic field captures sound generation and propagation.

Spatial uncertainty maps improve object localization success.

Explicit modeling enhances generalization over direct regression methods.

Abstract

We study the problem of multimodal physical scene understanding, where an embodied agent needs to find fallen objects by inferring object properties, direction, and distance of an impact sound source. Previous works adopt feed-forward neural networks to directly regress the variables from sound, leading to poor generalization and domain adaptation issues. In this paper, we illustrate that learning a disentangled model of acoustic formation, referred to as disentangled acoustic field (DAF), to capture the sound generation and propagation process, enables the embodied agent to construct a spatial uncertainty map over where the objects may have fallen. We demonstrate that our analysis-by-synthesis framework can jointly infer sound properties by explicitly decomposing and factorizing the latent space of the disentangled model. We further show that the spatial uncertainty map can significantly improve the success rate for the localization of fallen objects by proposing multiple plausible exploration locations.