Scholarlink

<p>Accurate state estimation is essential for reliable path-following control of unmanned surface vehicles (USVs), especially when low-cost sensors and nonlinear maneuvers are involved. In small USVs, heading and position measurements obtained from magnetic compasses and single GNSS modules are often affected by noise and disturbances, which degrade estimation accuracy in closed-loop control systems. To address this problem, this study establishes a unified closed-loop framework integrating line-of-sight (LOS) guidance, heading control, and Bayesian state estimation. First, a high-fidelity USV dynamic model is developed to simulate realistic nonlinear maneuvering behaviors. Second, multiple filtering algorithms, including KF, EKF, UKF, CKF3, and CKF5, are systematically implemented under compass-based and GNSS-based measurement schemes. Finally, Monte Carlo simulations are conducted in different maneuvering scenarios-straight-line motion, waypoint switching, and constant rudder turning—to evaluate estimation accuracy and computational efficiency. The results show that KF and EKF perform well under weak nonlinear conditions but degrade in strongly nonlinear maneuvers. UKF and CKF methods exhibit stronger robustness, with CKF5 achieving the highest accuracy, while CKF3 and UKF provide a better balance between accuracy and computational cost. This study provides practical guidelines for filter selection in USV path-following control.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Kalman Filter-based Path Following Control Design for Unmanned Surface Vehicles

  • 期刊论文

  • 发表日期：

    2026年5月13日

查看全文
  • Tengfei Sun,
  • Zhuoyi Yang,
  • Mingzhi Shao,
  • Lei Song,
  • Xin Liu,
  • Shufeng Feng

摘要

Accurate state estimation is essential for reliable path-following control of unmanned surface vehicles (USVs), especially when low-cost sensors and nonlinear maneuvers are involved. In small USVs, heading and position measurements obtained from magnetic compasses and single GNSS modules are often affected by noise and disturbances, which degrade estimation accuracy in closed-loop control systems. To address this problem, this study establishes a unified closed-loop framework integrating line-of-sight (LOS) guidance, heading control, and Bayesian state estimation. First, a high-fidelity USV dynamic model is developed to simulate realistic nonlinear maneuvering behaviors. Second, multiple filtering algorithms, including KF, EKF, UKF, CKF3, and CKF5, are systematically implemented under compass-based and GNSS-based measurement schemes. Finally, Monte Carlo simulations are conducted in different maneuvering scenarios-straight-line motion, waypoint switching, and constant rudder turning—to evaluate estimation accuracy and computational efficiency. The results show that KF and EKF perform well under weak nonlinear conditions but degrade in strongly nonlinear maneuvers. UKF and CKF methods exhibit stronger robustness, with CKF5 achieving the highest accuracy, while CKF3 and UKF provide a better balance between accuracy and computational cost. This study provides practical guidelines for filter selection in USV path-following control.