Transition Maneuver Optimization of a Hybrid UAV Using Deep Reinforcement Learning

Akhtar M., Maqsood A.

20th International Bhurban Conference on Applied Sciences and Technology (IBCAST), Bhurban, Pakistan, 22 - 25 Ağustos 2023, ss.214-219, (Tam Metin Bildiri) identifier identifier

  • Yayın Türü: Bildiri / Tam Metin Bildiri
  • Doi Numarası: 10.1109/ibcast59916.2023.10712871
  • Basıldığı Şehir: Bhurban
  • Basıldığı Ülke: Pakistan
  • Sayfa Sayıları: ss.214-219
  • Orta Doğu Teknik Üniversitesi Kuzey Kıbrıs Kampüsü Adresli: Hayır

Özet

In the past few years, the increasing need for dynamic mission profiles of UAVs has led to the development of autonomous aerial vehicles with sophisticated path planning, tracking, and control algorithms. Owing to the size and aerodynamic structure of small UAVs, they can achieve maneuvers like VTOL (Vertical Take off and Landing) and transition maneuvers from hover to cruise mode. Trajectory optimization of such maneuvers helps to minimize flight time and path while considering spatial constraints. The emergence of Machine Learning has gained attention for developing control systems based on such algorithms. This study mainly explores the algorithm Deep Deterministic Policy Gradient (DDPG) for path planning and optimization. To enhance the capabilities of small UAVs, a hybrid tiltrotor aircraft model is considered, which aims to hover and transition to an efficient forward flight during the cruise.