Robust Deep Multi-Modal Sensor Fusion using Fusion Weight Regularization and Target Learning

TL;DR

This paper introduces a robust deep multi-modal sensor fusion approach that uses fusion weight regularization and target learning to improve resilience against sensor failures, outperforming existing methods especially with corrupted inputs.

Contribution

The paper presents novel deep multi-modal sensor fusion architectures with fusion weight regularization and target learning, enhancing robustness under sensor failure conditions.

Findings

Outperforms existing architectures on clean and corrupted data

Significantly improves robustness when multiple sensors are corrupted

Demonstrates effectiveness in real-world sensor failure scenarios

Abstract

Sensor fusion has wide applications in many domains including health care and autonomous systems. While the advent of deep learning has enabled promising multi-modal fusion of high-level features and end-to-end sensor fusion solutions, existing deep learning based sensor fusion techniques including deep gating architectures are not always resilient, leading to the issue of fusion weight inconsistency. We propose deep multi-modal sensor fusion architectures with enhanced robustness particularly under the presence of sensor failures. At the core of our gating architectures are fusion weight regularization and fusion target learning operating on auxiliary unimodal sensing networks appended to the main fusion model. The proposed regularized gating architectures outperform the existing deep learning architectures with and without gating under both clean and corrupted sensory inputs resulted from sensor failures. The demonstrated improvements are particularly pronounced when one or more multiple sensory modalities are corrupted.