Sensor-based Autonomous Path-Planner for Sea-Bottom Exploration and Mosaicking∗

A typical mission for an underwater vehicle is to research and map OOIs (Objects Of Interest). Objects Of Interest are specified in connection with the mission purposes, relevant examples span from biological species (e.g. seagrass or coral colonies) to artifacts (e.g. ship wrecks). For their research and mapping the deployment of Autonomous and Remotely Operated Vehicles is becoming the state-of-art, due to their ease of performing difficult operations in a safe way. Moreover, photo-mosaicking tools can be used to provide a complete map of the explored area. In order to achieve this aim, the navigation usually relies on simple lawn-mower or zig-zag patterns, where an area is scanned extensively to collect the pictures, that would be analyzed, processed and stitched after the mission. This paper aims to increase the autonomy of underwater vehicles that are equipped with a still-picture camera by using a deliberative, sensor-driven planning strategy. By sensing the environment extracting its information from the pictures, the path can be defined on-the-fly, in order to increase the collected data.