Information Fusion Filtering Algorithm for Motion Target Tracking

This document presents an information fusion filtering algorithm based on the Interacting Multiple Model (IMM) framework, designed to track single moving targets exhibiting constant velocity-acceleration-constant velocity motion patterns. The algorithm integrates data from multiple sensors to achieve more accurate trajectory prediction and tracking. Key implementation involves maintaining multiple Kalman filters corresponding to different motion models, with probabilistic weight adjustments based on model likelihood calculations. The system automatically adapts to target motion states through model transition probabilities and selects optimal tracking strategies using Bayesian inference, significantly improving tracking accuracy and stability. Additionally, the algorithm's modular structure allows for enhanced performance through incorporation of additional sensor data or fine-tuning of filtering parameters (such as process noise covariance and measurement noise matrices), making it adaptable to various tracking scenarios. Code implementation typically involves parallel filter execution, residual calculation for model probability updates, and state estimation fusion using weighted sum approaches.