通過機載單目相機與基於純角度信息編隊控制方法實現空中繩索拖曳系統的平衡

Abstract

在許多產業中，飛行載具利用繩索來搬運重物已經行之有年。近年來，由多無人飛行載具與多繩索所構成的空中繩索拖曳系統也被提出來共同搬運重物。因為空中繩索拖曳系統具有高承載量與更好的系統擴展性，這樣的系統也吸引了越來越多的研究。 本文特別專注在研究如何只藉由機載感測器來平衡整個空中繩索拖曳系統。與其他的研究比較起來，本文致力於展示空中繩索拖曳系統也可以在不依賴於外部動作捕捉系統就可以進行平衡。 為此，本論文採用了一套藉由機載單目相機來量測純角度訊息的編隊控制方法。挑選基於純角度訊息的方法而不是基於純距離的方法是因為相對角度是可以被簡單的感測器諸如單目相機來量測，而單目相機已經被證實是可以安裝在小型的無人飛行載具上。 更具體的來說，首先，本文會先探討載物平衡問題與編隊控制問題之間的關 係，然後載物平衡問題會被轉化成一個隊形維持問題。接著，在本文中總共有三 套基於純角度訊息的編隊控制方法會被應用來協調多台無人飛行載具的運動以維 持這個理想的隊形。然而這個編隊控制方法會需要這個隊形中的無人機相對姿態 訊息才能運作。因此，論文採用了一套現有的基於單目相機的視覺相對姿態估測 方法來估測此隊形中的相對姿態。另外，另一套自行設計的視野控制器會被應用來確保此視覺相對姿態估測方法的順利運作。 最後，整個方法的可行性與有效性會被大量的實際飛行實驗來做驗證。

For many years, the aerial vehicles have been used to transport payload via cable in many industries. In the recent years, the aerial cable towed system, which is composed of multiple unmanned aerial vehicles and multiple cables, have also been proposed to transport a payload cooperatively. Since the aerial cable towed system possesses the ad- vantages such as higher load capacity and better system scalability, the aerial cable towed system is attracting more and more researches. In particular, this thesis is focusing on how to stabilize the aerial cable towed system with only onboard sensor. Compared with other researches, this thesis is devoted to show that the aerial cable towed system can be stabilized without relying on external motion capture system. To this extent, a bearing-based formation control method via onboard monocular camera is applied. The reason for choosing the bearing-based method instead of distance- based method lies in the fact that the relative bearing can be measured with simpler sensor such as monocular camera, which is proved to be able to installed onboard the unmanned aerial vehicles. To be more concrete, the relation between the payload stabilization and the formation control is firstly discussed, and the payload stabilization problem is transformed into a formation maintaining problem. After that, in total three different kinds of bearing-based formation controller method are applied to coordinate the motion of multiple drones to maintain the desired formation. It is worth noted that for the bearing-based formation controller to work, the information of the relative pose between drones is needed. In this thesis, an existing relative pose state estimation algorithm based on only monocular camera is applied to estimate such information. In addition, another self designed field of view controller is applied to ensure the feasibility of the vision-based estimation algorithm. Finally, considerable real flight experiments are executed to show the feasibility and effectiveness of the method to stabilize an aerial cable towed system.