Probabilistic Minefield Intelligence: Integrating UAV Detection with Statistical Combat Modeling

Abstract

This study presents a combined approach for mine- field intelligence by using UAV-based landmine detection together with statistical combat modelling. In this work, hyperspectral data from UAV sensors is processed through methods like Spec- tral Angle Mapper (SAM) and Adaptive Cosine Estimator (ACE) to create detection-probability maps. These detection scores are then merged with a probabilistic encounter model that uses a Hy- pergeometric formulation to estimate the chance of mine–target interaction under different mine densities and terrain conditions. The model also includes kill probability, cumulative mine activa- tion during movement, and terrain-based lethality adjustments. Along with the analytical model, a step-by-step simulation engine is developed to produce encounter patterns, attrition results, and sensitivity behaviour. Monte Carlo runs, variance checks, and confidence-interval analysis show good agreement between the UAV-based probability maps and the theoretical predictions. Visual outputs such as heatmaps and surface plots help to understand how encounter probability changes with mine density, detection confidence, and target shape. Overall, this framework offers a practical and statistically sound method for joining UAV reconnaissance with probabilistic combat modelling. It can support applications in detection evaluation, safe-route planning, sensor improvement, and future machine-learning-based mine- field assessment.

Keywords