First Trial of Underwater Excavator Work Supported by Acoustic Video Camera

抄録

External recognition is important for underwater machinery works. However, acquisition of external field information from optical camera images may not be possible, owing to muddiness of water caused by such work. Furthermore, in order to improve the workability of machines in the scenario of their remote operation, it is important to know the positional relation information between a target object and the end effector. To solve these problems, an acoustic video camera was developed and performance test experiments were conducted at a caisson dockyard. In the experiments, a prototype of acoustic video camera was used to measure and to recognize a target objects and an underwater construction machine. And the feasibility of monitoring for underwater construction using the acoustic videos was evaluated. As a result, it was found that despite the lower accuracy of shape recognition on account of a resolution problem, the positional relation could be recognized satisfactorily since the video images could be presented from an arbitrary viewpoint.