Autonomous recognition of bubble plumes for navigation of underwater robots in active shallow vent areas

Although underwater vent fields are of great scientifc interest, little is understood concerning their influence on the underwater ecosystem due to the great difficulties involved in their underwater survey. The authors propose an underwater survey method using AUVs. In previous works, the authors proposed a practical method to survey the "Tagiri" vent area in Kagoshima, in southern Japan [1] [2], and carried out experiments. The field experiments in Tagiri were at some degree successful, but the system also encountered some difficulties. One of the problems was that the AUV could not identify the type of landmarks used to support accurate positioning. In this paper, we propose an autonomous recognition method of bubble plumes using a sheet laser and a camera by image processing. The performance of the proposed method was verified with data obtained in tank experiments, and then implemented in the AUV Tri-Dog 1. Using the proposed method, Tri-Dog 1 performed seafloor observations of the Tagiri vent area in March 2007. Tri-Dog 1 succeeded in distinguishing between bubble plumes and artificial acoustic reflectors, and localizing itself with sufficient accuracy to create a vast, high resolution visual map of the area.