Robust Inference for Visual-Inertial Sensor Fusion

TL;DR

This paper develops robust filtering methods for visual-inertial sensor fusion that effectively handle outliers, improving state estimation accuracy without increasing computational costs.

Contribution

It derives the optimal discriminant for outlier detection and introduces new approximation methods, enhancing robustness in visual-inertial fusion systems.

Findings

Best method improves state-of-the-art performance

Maintains same computational complexity

Analytical and empirical comparison of methods

Abstract

Inference of three-dimensional motion from the fusion of inertial and visual sensory data has to contend with the preponderance of outliers in the latter. Robust filtering deals with the joint inference and classification task of selecting which data fits the model, and estimating its state. We derive the optimal discriminant and propose several approximations, some used in the literature, others new. We compare them analytically, by pointing to the assumptions underlying their approximations, and empirically. We show that the best performing method improves the performance of state-of-the-art visual-inertial sensor fusion systems, while retaining the same computational complexity.