The Implementation of the SLAM Method in ROS System and Underwater Robot

Abstract

Autonomous localization and map construction of underwater robots in unknown environments is a very important research area. Visual SLAM technology is based on visual processing algorithms and can accomplish this task at a low cost. This paper focuses on a set of underwater robot visual SLAM experimental process. Based on ROS system, Mono visual SLAM experiments are performed on UWSim simulation environment and real robot respectively by using mature ORB-SLAM2 algorithm. The experimental results show that this method can accurately build the sparse point cloud map of the environment under the condition of sufficient environmental feature points.