Skip to main navigation Skip to search Skip to main content

Toward Reinforcement Learning-Based Filter Switching for Radar Target Tracking

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Abstract

We propose an adaptive filter-switching framework for radar target tracking, integrating reinforcement learning to intelligently select among multiple Kalman filter models based on target dynamics. Specifically, a Deep Q-Network agent is trained to choose the most suitable filter at each timestep to minimize tracking error. The framework is evaluated on a target trajectory with distinct motion phases and maneuvers. Performance is benchmarked against the Kalman Filters used and an Interacting Multiple Model estimator. Results show that the DQN-based filter switcher achieves comparable accuracy, demonstrating the viability of learned filter selection for real-time tracking applications.

Original languageEnglish (US)
Title of host publicationNAECON 2025 - IEEE National Aerospace and Electronics Conference
PublisherInstitute of Electrical and Electronics Engineers Inc.
ISBN (Electronic)9798331538132
DOIs
StatePublished - 2025
Event2025 IEEE National Aerospace and Electronics Conference, NAECON 2025 - Dayton, United States
Duration: Jul 28 2025 → Jul 31 2025

Publication series

NameProceedings of the IEEE National Aerospace Electronics Conference, NAECON
ISSN (Print)0547-3578
ISSN (Electronic)2379-2027

Conference

Conference2025 IEEE National Aerospace and Electronics Conference, NAECON 2025
Country/TerritoryUnited States
CityDayton
Period7/28/25 → 7/31/25

All Science Journal Classification (ASJC) codes

  • Control and Systems Engineering
  • Computer Science Applications
  • Computer Networks and Communications
  • Electrical and Electronic Engineering

Fingerprint

Dive into the research topics of 'Toward Reinforcement Learning-Based Filter Switching for Radar Target Tracking'. Together they form a unique fingerprint.

Cite this