Accurate navigation is a critical requirement for many unmanned underwater vehicles missions. This becomes even more apparent when operating in near intervention and hazardous locations such as near the seabed. The use of vision systems and image processing as a means of high precision horizontal motion estimation in near seabed missions can bring many advantages over commercially available sensor solutions. This paper discusses the vision system and the data collection method used for vision based navigation investigation on the newly developed smart remotely operated vehicle, ROVLATIS. ROV LATIS has undergone full sea trials onboard the RV Celtic Explorer off the west coast of Ireland. The key results and conclusions from these trials are presented. © 2009 IFAC.