Design and Flight Test of a Quality Requirements-Based Controller for the Skydolphin Aerial Robot


Çelik H.

2026 International Conference on Advanced Robotics and Mechatronics (ICARM), Paris, Fransa, 31 Temmuz - 06 Ağustos 2026, ss.174-179, (Tam Metin Bildiri)

  • Yayın Türü: Bildiri / Tam Metin Bildiri
  • Doi Numarası: 10.1109/icarm69740.2026.11774757
  • Basıldığı Şehir: Paris
  • Basıldığı Ülke: Fransa
  • Sayfa Sayıları: ss.174-179
  • Erciyes Üniversitesi Adresli: Evet

Özet

A fundamental aspect of designing conventional, aft-tailed, fixed-wing aerial robots is achieving an inherently stable airframe. However, establishing static stability-the vehicle's initial tendency to return to its original equilibrium state after being disturbed-does not guarantee dynamic stability. Despite designers' efforts, achieving dynamic stability may not be feasible due to various constraints, including performance requirements and geometric limitations. After establishing an inherently statically stable design, meeting dynamic stability requirements may necessitate updating the airframe. Since updating a statically stable airframe is challenging, an alternative approach is to implement stability augmentation or control systems to achieve dynamic stability, as represented by flying quality requirements. This study aims to reveal the flight dynamics characteristics of the SkyDolphin, a laboratory-produced aerial robot, and satisfy the desired flying quality requirements through controller design. First, the aerodynamics and flight dynamics models of the SkyDolphin are derived, and its flying quality characteristics are evaluated. This evaluation reveals the need for a flight control system to ensure dynamic stability. Next, a flight control system based on these requirements was designed and tested in a simulation and real environments. Finally, the results of the experimental evaluation of the controller performance and the robot's flying quality in a real flight environment are presented. Developing the baseline controller provided a testbed for evaluating and comparing the performance of future reinforcement learning controller counterparts in satisfying flying qualities during real-time flights.