Viewpoint Generation for Coverage Inspection of Airplane Fuselage Surface Based on Unmanned Ground Vehicles

Traditional manual visual inspection of the airplane surface has the disadvantages of low efficiency, poor safety, and incapability of full coverage. To address these problems, the unmanned ground vehicle (UGV) is employed to perform the coverage inspection automatically. Viewpoint generation for the UGV, which is the key to high coverage rate and quality, is the main concern of this article. To improve the coverage quality, photogrammetric requirements including the equidistance, front view, and overlap are introduced. Furthermore, many practical constraints, such as the safe distance, height and angles of the lifting rod and pan-tilt unit, are also considered. To satisfy the above requirements and constraints, we propose a mixed sampling method that combines the dual sampling and the primal sampling. It is worth noting that the computational complexity can achieve the near real-time performance. Then, the coverage produced by the viewpoints is evaluated by using the frustum and culling occlusion approach. Finally, calculations on a Boeing 737–300 are conducted, and the results demonstrate the effectiveness of our method.