Hierarchical Bayesian Data Fusion for Robotic Platform Navigation

TL;DR

This paper introduces a hierarchical Bayesian data fusion method for robotic navigation that combines multiple sensor measurements adaptively, outperforming individual trackers in simulated and real robotic scenarios.

Contribution

A novel hierarchical Bayesian fusion framework that enhances sensor data integration for robotic navigation, addressing limitations of existing machine learning-based methods.

Findings

Outperforms individual trackers in simulated tests

Effective in real robotic platform experiments

Adaptive fusion improves robustness and accuracy

Abstract

Data fusion has become an active research topic in recent years. Growing computational performance has allowed the use of redundant sensors to measure a single phenomenon. While Bayesian fusion approaches are common in general applications, the computer vision field has largely relegated this approach. Most object following algorithms have gone towards pure machine learning fusion techniques that tend to lack flexibility. Consequently, a more general data fusion scheme is needed. Within this work, a hierarchical Bayesian fusion approach is proposed, which outperforms individual trackers by using redundant measurements. The adaptive framework is achieved by relying on each measurement's local statistics and a global softened majority voting. The proposed approach was validated in a simulated application and two robotic platforms.