Spatial constraints improve filtering of measurement noise from animal tracks

2025-11-27

AI Summary

To address the challenges of non-Gaussian, heavy-tailed measurement noise and insufficient modeling of geographic spatial constraints in animal tracking using GPS/Argos data, this paper proposes a latent-variable movement model explicitly incorporating spatial constraints. The animal’s dynamics are modeled via an underdamped Langevin stochastic differential equation, with a boundary-aware term explicitly embedded in the state evolution to enforce geographic feasibility. For inference, we hybridize the extended Kalman filter with a particle filter tailored to heavy-tailed observation errors, enabling robust state estimation under non-Gaussian noise. Experiments demonstrate that our method significantly improves trajectory denoising accuracy and robustness across diverse heavy-tailed noise regimes. Notably, on real-world Argos and GPS tracking data, it outperforms conventional Kalman-based approaches. This yields more reliable trajectory reconstructions, thereby strengthening downstream ecological niche modeling and behavioral analysis.

Technology Category

• Intelligent Robots: State Estimation
• Reasoning under Uncertainty: Stochastic Optimization
• Computer Vision: Motion & Tracking

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Security and Privacy: Large-scale security measurements

Abstract

Advances in tracking technologies for animal movement require new statistical tools to better exploit the increasing amount of data. Animal positions are usually calculated using the GPS or Argos satellite system and include potentially complex non-Gaussian and heavy-tailed measurement error patterns. Errors are usually handled through a Kalman filter algorithm, which can be sensitive to non-Gaussian error distributions. In this paper, we introduce a realistic latent movement model through an underdamped Langevin stochastic differential equation (SDE) that includes an additional drift term to ensure that the animal remains in a known spatial domain of interest. This can be applied to aquatic animals moving in water or terrestrial animals moving in a restricted zone delimited by fences or natural barriers. We demonstrate that the incorporation of these spatial constraints into the latent movement model improves the accuracy of filtering for noisy observations of the positions. We implement an Extended Kalman Filter as well as a particle filter adapted to non-Gaussian error distributions. Our filters are based on solving the SDE through splitting schemes to approximate the latent dynamic.

Problem

• Develops a latent movement model using SDEs to handle non-Gaussian measurement errors in animal tracking.
• Incorporates spatial constraints to improve filtering accuracy for noisy animal position data.
• Implements Extended Kalman and particle filters adapted for complex error distributions in movement data.

Innovation

• Using Langevin SDE with spatial drift constraints
• Implementing Extended Kalman and particle filters
• Applying splitting schemes to solve SDE dynamics