Autonomous underwater vehicle navigation and mapping in dynamic, unstructured environments

Abstract

This thesis presents a system for automatically building 3-D optical and bathymetric maps of underwater terrain using autonomous robots. The maps that are built improve the state of the art in resolution by an order of magnitude, while fusing bathymetric information from acoustic ranging sensors with visual texture captured by cameras. As part of the mapping process, several internal relationships between sensors are automatically calibrated, including the roll and pitch offsets of the velocity sensor, the attitude offset of the multibeam acoustic ranging sensor, and the full six-degree of freedom offset of the camera. The system uses pose graph optimization to simultaneously solve for the robot’s trajectory, the map, and the camera location in the robot’s frame, and takes into account the case where the terrain being mapped is drifting and rotating by estimating the orientation of the terrain at each time step in the robot’s trajectory. Relative pose constraints are introduced into the pose graph based on multibeam submap matching using depth image correlation, while landmark-based constraints are used in the graph where visual features are available. The two types of constraints work in concert in a single optimization, fusing information from both types of mapping sensors and yielding a texture-mapped 3-D mesh for visualization. The optimization framework also allows for the straightforward introduction of constraints provided by the particular suite of sensors available, so that the navigation and mapping system presented works under a variety of deployment scenarios, including the potential incorporation of external localization systems such as long-baseline acoustic networks. Results of using the system to map the draft of rotating Antarctic ice floes are presented, as are results fusing optical and range data of a coral reef.