自動離着桟システムに向けたステレオ画像による船舶の検出と位置姿勢推定に関する検討

抄録

  Most studies on autonomous ships are aimed at autonomous control in navigation, berthing and collision avoidance maneuvers on the premise that the existing marine equipment, such as radar, AIS and GPS systems are used to aid autonomous operation. However, these devices are not enough to acquire information about surrounding conditions in detail, which is very important for autonomous navigation. The authors have proposed a 3D measurement method using a stereo vision system that helps obtain various kinds of information about wider areas around the ship. In particular, the authors have been studying to estimate the position and attitude of a target ship with high accuracy by using cameras placed on land for automatic berthing systems. First, this paper discusses a method for automatically determining the optimum placement of stereo cameras to observe the target ship. Then it describes a method for detecting a ship from a disparity image using the concept of U-Disparity / V-Disparity images. Lastly, the paper illustrates that the position and attitude of a ship are estimated in a series of scenes in which the ship is approaching the berth, and indicates the effectiveness of the proposed method by comparing the results of this estimation with AIS data.