Direct estimation of motion from sea floor images for automatic station-keeping of submersible platforms

Abstract

Maintaining a fixed position near the sea floor is a critical capability during the deployment of remotely operated or intelligent (autonomous) undersea vehicles in a variety of missions, including inspection and repair of undersea structures, data collection, and surveillance. We present an automatic optical station-keeping system for application to submersible vehicles in deep waters by exploiting the information in sea floor images. Readily measurable spatio-temporal image gradients are used to detect and compute the vehicle's translational and yaw motions using a direct motion vision technique. The vision system has been implemented on a Windows-NT Pentium platform, and the estimated positions and yaw angles are communicated via a serial link to the control system, running on a PC-386. Accurate station-keeping is demonstrated in experiments with a three-thruster floating vehicle in a 6-ft/spl times/12-ft/spl times/6-ft water tank.