水中ロボットによる沈没船周辺のワールドモデルの自動構築

概要

論文の詳細を見る
For a close survey mission to a wreck such as the Russian tanker Nakhodka, the ROV is a powerful tool and provides us vivid images of the wreck. The ROV is controlled by human operators on the support vessel based on the visual information from the vehicle, knowledge of the ship, and operational know-how. Since the width of the visual field is limited, the operators may misunderstand the position of the vehicle. Moreover, the operation near the wreck is a significantly dangerous task and often causes the loss of the vehicle. This paper proposes a world modeling system to display whale's-eye view of the target area in real time to help the ROV operators to understand the surrounding environment of the vehicle. The ranging SONAR data are accumulated and processed to update the world model while the vehicle is operated along a path which is determined based on the model. Results of experiments in a testing pool and computer simulations represent that the proposed system can draw elevation maps and 3 D occupancy grid maps of the environment with complicated shapes of obstacles.