Statistical Motion Prediction of Rogue UAVs for Efficient Pursuit and Capture Using a Hybrid Kalman-Particle Filter Framework

The growing rogue or intruder aerial incursions demand a robust yet stable interception mechanism that can estimate the UAV state considering the measurement inaccuracies and process noise. To estimate, track and capture rogue aerial nodes, navigating through 2D way points with inherent Gaussian process noise, the study introduces a hybrid tracking and estimation framework, integrating Kalman and Particle filters. The integration facilitates tracking and predicting both linear and abrupt non linear maneuvers along the rogue trajectory. The rogue aerial motion is modeled through a random burst process using a random acceleration, while the interceptor UAV employs a pursuit-evasion control. The evasion control in turn is driven by the probabilistic state estimates received from the Hybrid filter. Accuracy, energy efficiency and interception success rate of the proposed framework is measured across extensive Monte Carlo simulations (10,000 trials), under varied noise intensities. Comparative analysis against single-filter methods suggest that the Hybrid filter improves trajectory, achieves better estimation accuracy and high interception probability. Under an unpredictable and dynamic aerial environment, the suggested framework allows for adaptive, real-time pursuit and capture of rogue UAVs.