Research on Vision-Based ROV Underwater Positioning Method

ROV has the advantages of multiple functions, strong comprehensive operation capabilities, safety and reliability, and real-time remote control. It is widely used in fields such as marine resource development and marine geological survey. Usually, the ROV operating scope covers regional environments such as underwater reefs, and there are large errors in sonar positioning information. This paper aims at the problem of inaccurate ROV motion positioning information in operating scenarios. Based on the ORB-SLAM visual positioning method, the L-M method in back-end optimization is improved to reduce the ROV pose drift error and achieve accurate ROV positioning in small scenes. By building the algorithm framework of this article and conducting pool experiments based on the ROV platform, we verified the feasibility of the algorithm and the positioning accuracy of the ROV.