Scholarlink

<p>Consequent to the previous work published by Boase&#xa0;et al., this paper presents a novel data-driven multi-input multi-output (MIMO) control strategy for real-time spatial trajectory tracking of lighter-than-air dirigible airships under external disturbances. The proposed control framework leverages online data collection along the generated trajectories to enable learning of the airship’s actuation mechanisms using a model-free reinforcement learning (RL) approach. Specifically, we extend our prior results by employing an actor-critic RL methodology formulated as an optimal control problem to achieve robust trajectory tracking. The effectiveness of the developed strategy is validated within a highly realistic ROS/Gazebo environment, demonstrating reliable performance across various environmental complexities. Additionally, the proposed online data-driven controller is benchmarked against a conventional model-based controller designed for similar airship applications.</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

Disturbance-resilient learning and control for dirigible airships with imprecise system dynamics

  • 期刊论文

  • 发表日期：

    2025年8月4日

查看全文
  • Derek Boase,
  • Md Suruz Miah,
  • Wail Gueaieb

摘要

Consequent to the previous work published by Boase et al., this paper presents a novel data-driven multi-input multi-output (MIMO) control strategy for real-time spatial trajectory tracking of lighter-than-air dirigible airships under external disturbances. The proposed control framework leverages online data collection along the generated trajectories to enable learning of the airship’s actuation mechanisms using a model-free reinforcement learning (RL) approach. Specifically, we extend our prior results by employing an actor-critic RL methodology formulated as an optimal control problem to achieve robust trajectory tracking. The effectiveness of the developed strategy is validated within a highly realistic ROS/Gazebo environment, demonstrating reliable performance across various environmental complexities. Additionally, the proposed online data-driven controller is benchmarked against a conventional model-based controller designed for similar airship applications.