Of heading, posture and body rotations derived from data acquired by animal-borne accelerometers, magnetometers and gyrometers, kernel density estimation of the corresponding spherical distributions, and fine-scale movement reconstruction

TL;DR

This paper develops mathematical tools to derive animal body orientations and movements from high-frequency sensor data, using spherical KDE for distribution analysis and addressing data management challenges.

Contribution

It introduces classical trigonometric formulas for 3D orientation and rotation computation from accelerometer, magnetometer, and gyrometer data, along with spherical KDE methods for movement analysis.

Findings

Formulated mathematical expressions for 3D orientation and rotation from sensor data.

Established methods for representing spherical distributions via KDE.

Provided approaches for managing large high-frequency datasets.

Abstract

This paper provides the mathematical expressions, formulated in classical trigonometric terms, that are required (i) to compute an animal's body orientations and rotations in 3D space from data acquired at high-frequency by on-board accelerometers, magnetometers and gyrometers, (ii) to relate 3D rotations to changes in 3D orientation, (iii) to represent their spherical distributions through kernel density estimation (KDE), and (iv) to reconstruct fine-scale movements. The way the huge amount of data that is acquired by these on-board devices can be more easily managed is also considered.