Short Range Tracking Method of Underwater Vehicles Based on Vision

In this paper, the authors propose a method for an autonomous underwater vehicle (AUV) to track an underwater vehicle at short distances using a vision-based approach where the AUV estimates self-position and velocity with regards to a target vehicle. AUVs have been commonly used to explore the seafloor. However, since the observation coverage of a single AUV is limited, survey is using multiple vehicles has been proposed. An acoustic or visual approach is usually used for positioning and communication between vehicles. Although an acoustic approach is suitable for long distance positioning and communication data rates are very low. For more frequent and accurate positioning at short distances, the authors adopted visual positioning. In the proposed method, an AUV estimates its position and velocity at short range against a moving vehicle using its camera and an LED array mounted on the other vehicle. The novelty of the proposed method is that the AUV estimates its relative velocity against the moving vehicle and utilizes it for self-control. Since the proposed method can be implemented with only a monocular camera and LEDs, it can be easily applied to low-cost vehicles. The proposed method can also be effective in automatic recovery of an AUV to a vessel. To evaluate the performance of the proposed method, A tank test was conducted using an AUV.