A Low-Cost Visual Inertial Odometry System for Underwater Vehicles

The localization is a crucial issue for underwater vehicles. In the paper, a lightweight visual-inertial odometry is proposed. With dual inertial sensors giving the information of acceleration and attitude, an optical camera providing the seabed images where feature points are tracked by an optical flow algorithm, linear motion of the vehicle can be obtained by computing coordinate transformations and in the fusion section, the control input is also considered. The computational complexity of the proposed method is reduced dramatically relative to other methodologies, and the optical flow algorithm can allow the system to work in poor context environment conditions. The results evaluated by practical experiments show that the method is an effective, low-cost solution for underwater localization.