A Comparative Study in Tracking Highly-Maneuvrable Targets Using Fuzzy Logic Particle Filter and Interacting Multiple Models

A highly-maneuverable target may display variable rates of acceleration, sharp maneuvers, thrust, and rotation, deviating from a strictly linear behavior. While the Interacting Multiple Models (IMM) strategy, known for its multimodal nature, necessitates familiarity with the target model in advance. To overcome this restriction, a Fuzzy Logic Particle Filter (FLPF) is employed. The FLPF utilizes fuzzy relational equations in a single-input single-output configuration, presented in a rule-based canonical structure. The complex movements of the maneuvering target are depicted through a series of switching systems. Unlike IMM, the FLPF does not call for statistical model information and does not mandate a maneuver sensor to track maneuvering targets, resulting in reduced computational complexity. Through the use of a fuzzy overlap set, only a subset of models require evaluation based on similar acceleration approximations, diminishing computational demands in contrast to IMM. The incorporation of additional models does not escalate computational requirements, as the quantity of models assessed is dictated by the degree of overlap. An illustrative case showcases the efficiency of the proposed technique, with simulations demonstrating that the FLPF offers effective tracking capabilities with decreased computational strain when compared to IMM.