Year: 2025
Type: conference-paper
Abstract: The Box Particle Filter (Box-PF) is an improved algorithm of the Particle Filter (PF) and a generalized particle filtering algorithm for dealing with uncertain measurements. It has attracted considerable attention due to its use of box particles and error bounds models in Sequential Monte Carlo methods, replacing the traditional point particles and error statistical models. In target tracking systems, the Box-PF algorithm suffers from complex logic and a large amount of data in the interval analysis step, which negatively impacts both the real-time performance and tracking accuracy of the algorithm. To address these issues, this study proposes an improved Box-PF algorithm for target tracking. The improved algorithm enhances the mathematical representation of box particles and the boundaries of the state space by using spherical shells. It eliminates the step of calculating the minimum bounding function and simplifies the contraction and weight update steps of the algorithm. These improvements enhance the tracking accuracy and reduce the computational complexity of the algorithm. Simulation experiments show that the Spherical Shell Box-PF algorithm achieves a 21.07% improvement in tracking accuracy and a 74.12% reduction in running time compared to the traditional Box-PF algorithm.
Authors: Yifeng Qiu, Daqing Huang
Institution: Nanjing University of Aeronautics and Astronautics
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W4411539674