MIXER: A Principled Framework for Multimodal, Multiway Data Association

TL;DR

This paper introduces MIXER, a new framework that improves multimodal, multiway data association in robotic perception by leveraging a mixed-integer quadratic approach with a novel relaxation, leading to more stable and accurate correspondences.

Contribution

The paper presents a principled mixed-integer quadratic framework with a novel continuous relaxation for multimodal, multiway data association, enhancing robustness and accuracy.

Findings

35% increase in F1 score on robotics dataset

More stable correspondences under noise and errors

Outperforms state-of-the-art techniques

Abstract

A fundamental problem in robotic perception is matching identical objects or data, with applications such as loop closure detection, place recognition, object tracking, and map fusion. While the problem becomes considerably more challenging when matching should be done jointly across multiple, multimodal sets of data, the robustness and accuracy of matching in the presence of noise and outliers can be greatly improved in this setting. At present, multimodal techniques do not leverage multiway information, and multiway techniques do not incorporate different modalities, leading to inferior results. In contrast, we present a principled mixed-integer quadratic framework to address this issue. We use a novel continuous relaxation in a projected gradient descent algorithm that guarantees feasible solutions of the integer program are obtained efficiently. We demonstrate experimentally that correspondences obtained from our approach are more stable to noise and errors than state-of-the-art techniques. Tested on a robotics dataset, our algorithm resulted in a 35% increase in F1 score when compared to the best alternative.