日本フィジカルAI新聞

世界のフィジカルAIを、日本語で。

制御/適応制御arXiv:1907.08619

自己適応型弾性構造コントローラによるUAVのリアルタイム複雑ミッション

Real-time UAV Complex Missions Leveraging Self-Adaptive Controller with Elastic Structure

シェア:XThreadsFacebookLINEはてブBluesky

スライディングモード制御と進化型ニューロファジィ制御を組み合わせ、ファジィルール数を適応的に増減させる弾性構造の自己適応コントローラを四 rotor UAVに実装し、地面効果・天井効果・強風下での軌道追従を実機評価した。

著者: Mohamad Abdul Hady, Basaran Bahadir Kocer, Harikumar Kandath, Mahardhika Pratama

分類: eess.SY, cs.RO, cs.SY

原文アブストラクト

The expectation of unmanned air vehicles (UAVs) pushes the operation environment to narrow spaces, where the systems may fly very close to an object and perform an interaction. This phase brings the variation in UAV dynamics: thrust and drag coefficient of the propellers might change under different proximity. At the same time, UAVs may need to operate under external disturbances to follow time-based trajectories. Under these challenging conditions, a standard controller approach may not handle all missions with a fixed structure, where there may be a need to adjust its parameters for each different case. With these motivations, practical implementation and evaluation of an autonomous controller applied to a quadrotor UAV are proposed in this work. A self-adaptive controller based on a composite control scheme where a combination of sliding mode control (SMC) and evolving neuro-fuzzy control is used. The parameter vector of the neuro-fuzzy controller is updated adaptively based on the sliding surface of the SMC. The autonomous controller possesses a new elastic structure, where the number of fuzzy rules keeps growing or get pruned based on bias and variance balance. The interaction of the UAV is experimentally evaluated in real time considering the ground effect, ceiling effect and flight through a strong fan-generated wind while following time-based trajectories.

関連論文

  1. 1Boston Dynamics、車輪付きAtlasの研究試作機を紹介研究・技術
  2. 2GXOとApptronik、Apolloの物流現場試験を拡大 最大1,000台の展開を構想量産・展開
  3. 3newmoとKDDIが資本業務提携 2028年に堺市で自動運転タクシー商用化目指す提携・買収
  4. 4NVIDIA、CoRLでGR00T 2とIsaac Lab 3.0を発表新製品・発表
  5. 5UC BerkeleyとNVIDIAの研究者、AIエージェントでロボット制御コードを仮想環境で改善研究・技術

PR・アフィリエイト広告

FRONTIER(GPU搭載BTOパソコン)

FRONTIER(GPU搭載BTOパソコン)の公式広告バナーFRONTIERでGPU搭載PCを見る ↗
PR本紙発行元 EmplifAI