Variable Structure IMM Algorithm - Adaptive Model-Set Switching IMM Algorithm (AGIMM)

This article explores the Variable Structure Interacting Multiple Model (IMM) algorithm, which is widely adopted for adaptive model-set switching due to its enhanced capability in handling uncertainty and dynamic changes. We also examine an alternative approach—the Adaptive Model-Set Switching IMM algorithm (AGIMM)—which has demonstrated superior performance in certain scenarios compared to the standard Variable Structure IMM. To facilitate optimal algorithm selection for specific applications, we provide a comparative analysis of both methods, highlighting their strengths and limitations. Key implementation aspects include model-set transition logic, probability update mechanisms, and filter fusion techniques. The AGIMM algorithm particularly emphasizes adaptive model activation/deactivation strategies through real-time likelihood evaluation, often implemented via threshold-based switching in practical codebases. Understanding these algorithmic differences enables better customization for tracking systems, sensor fusion, and other estimation tasks requiring dynamic model adaptation.