Automatic Docking Based on Visual Feedback on Remotely Operated Vehicle

摘要

Corals are defenders of destructive waves, home for fishes, and also a tourist attraction. In a past few years, corals lost their colors due to global warming. The development of robotics makes it possible to aid coral rehabilitation utilizing ROV (Remotely Operated Vehicle). This paper presents an automatic docking method for ROV using a camera as feedback. A marker is placed under a docking station. ROV estimates its position and orientation relative to this marker. PID control is then used to maintain its position and orientation to the marker. The system is applied on Raspberry Pi 3. The marker detection algorithm takes 0.2 s in each iteration on this platform. The rise time of position control is 20 s and 25 s for orientation control. ROV approaches the marker at 0.66 cm/s in a calm water condition. The docking system is proven to be accurate with 4.7% of position error and 11% of orientation error. The algorithm is proven to be applicable on a small computer such as Raspberry Pi.