Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

From Forecast to Action: Uncertainty-Aware UAV Deployment for Ocean Drifter Recovery

Jingeun Kim, Yong-Hyuk Kim, Yourim Yoon

TL;DR

This work presents an end-to-end predict-then-optimize framework for maritime search-and-rescue that couples LLM-based drifter trajectory prediction with uncertainty-aware UAV deployment optimization. By generating Gaussian particle samples around Chronos-predicted positions and dynamically adjusting UAV detection radii, the approach accounts for prediction error while seeking maximal coverage of potential trajectories via meta-heuristics (GA, PSO, SA) plus a repair mechanism. Real-world data from the Korean coast demonstrate the framework's robustness and superiority over random search, with consistent improvements across multiple model variants and instance settings. The study offers a practical blueprint for adaptive, uncertainty-aware resource deployment in time-critical SAR and other spatio-temporal domains.

Abstract

We present a novel predict-then-optimize framework for maritime search operations that integrates trajectory forecasting with UAV deployment optimization-an end-to-end approach not addressed in prior work. A large language model predicts the drifter's trajectory, and spatial uncertainty is modeled using Gaussian-based particle sampling. Unlike traditional static deployment methods, we dynamically adapt UAV detection radii based on distance and optimize their placement using meta-heuristic algorithms. Experiments on real-world data from the Korean coastline demonstrate that our method, particularly the repair mechanism designed for this problem, significantly outperforms the random search baselines. This work introduces a practical and robust integration of trajectory prediction and spatial optimization for intelligent maritime rescue.

From Forecast to Action: Uncertainty-Aware UAV Deployment for Ocean Drifter Recovery

TL;DR

This work presents an end-to-end predict-then-optimize framework for maritime search-and-rescue that couples LLM-based drifter trajectory prediction with uncertainty-aware UAV deployment optimization. By generating Gaussian particle samples around Chronos-predicted positions and dynamically adjusting UAV detection radii, the approach accounts for prediction error while seeking maximal coverage of potential trajectories via meta-heuristics (GA, PSO, SA) plus a repair mechanism. Real-world data from the Korean coast demonstrate the framework's robustness and superiority over random search, with consistent improvements across multiple model variants and instance settings. The study offers a practical blueprint for adaptive, uncertainty-aware resource deployment in time-critical SAR and other spatio-temporal domains.

Abstract

We present a novel predict-then-optimize framework for maritime search operations that integrates trajectory forecasting with UAV deployment optimization-an end-to-end approach not addressed in prior work. A large language model predicts the drifter's trajectory, and spatial uncertainty is modeled using Gaussian-based particle sampling. Unlike traditional static deployment methods, we dynamically adapt UAV detection radii based on distance and optimize their placement using meta-heuristic algorithms. Experiments on real-world data from the Korean coastline demonstrate that our method, particularly the repair mechanism designed for this problem, significantly outperforms the random search baselines. This work introduces a practical and robust integration of trajectory prediction and spatial optimization for intelligent maritime rescue.

Paper Structure

This paper contains 26 sections, 6 equations, 5 figures, 5 tables, 1 algorithm.

Table of Contents

  1. INTRODUCTION
  2. PROPOSED APPROACH
  3. Problem Definition
  4. Problem Modeling
  5. Search Area Definition
  6. Dynamic Detection Radius
  7. Probability of Detection (PoD) Model
  8. Prediction
  9. Optimization
  10. Initialization
  11. Genetic Operator
  12. Particle Swarm Optimization
  13. Simulated Annealing
  14. Repair
  15. Evaluation during Meta-heuristic algorithms
  16. ...and 11 more sections

Figures (5)

  • Figure 1: The proposed framework combining LLM-based trajectory prediction with UAV deployment optimization via meta-heuristic algorithms
  • Figure 2: Illustration of the minimum (200m) and maximum (600m) effective detection radii based on the number of spiral search revolutions.
  • Figure 3: Repair mechanism for UAV deployment
  • Figure 4: Visualization of the datasets used in this study. Red dots indicate accident occurrence locations on the map.
  • Figure 5: Performance comparison of GA, PSO, SA, and random search on Instance 1 with 6 UAVs and 10 particles across Chronos variations