Scholarlink

This paper presents a physics-data hybrid control framework for high-speed trajectory tracking of unmanned aerial vehicles (UAVs). The proposed methodology integrates a hybrid model, which comprises both physical dynamics and residual components learned via Gaussian Process (GP) regression, within a Model Predictive Control (MPC) framework. The GP model compensates for unmodeled aerodynamics disturbances while providing uncertainty estimates. MPC constructs a unified optimization framework by formulating a performance index that integrates both trajectory tracking errors and control input variation rates. The design enhances robustness and tracking accuracy under the high-speed and uncertain flight conditions. Simulations validate the potential of the framework for real-time agile UAVs.

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

Physics-Data Hybrid Control for High-Speed Trajectory Tracking of UAVs

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Yulin Huang,
  • Jiahao Wang,
  • Haotian Lou,
  • Yanbing Chen,
  • Shaojie Chen,
  • Mingkai Guo,
  • Jianxiao Zou,
  • Ping Liu

摘要

This paper presents a physics-data hybrid control framework for high-speed trajectory tracking of unmanned aerial vehicles (UAVs). The proposed methodology integrates a hybrid model, which comprises both physical dynamics and residual components learned via Gaussian Process (GP) regression, within a Model Predictive Control (MPC) framework. The GP model compensates for unmodeled aerodynamics disturbances while providing uncertainty estimates. MPC constructs a unified optimization framework by formulating a performance index that integrates both trajectory tracking errors and control input variation rates. The design enhances robustness and tracking accuracy under the high-speed and uncertain flight conditions. Simulations validate the potential of the framework for real-time agile UAVs.