Eleventh Coffee with EUMR

Access to the deep-sea is still limited by the availability of ship time and comes at a hefty cost. Testing new concepts or developing new systems for deep-sea research is non-trivial, adding complexity to the already challenging environmental and operational constraints of doing so in the upper 500m of the water column. To address this issue have been developing a new concept that is leveraging existing, proven infrastructure to gain access for smaller, task specific systems. In this talk this concept will be presented, providing a brief overview of the MARUM research infrastructure, we are going to capitalize on. Furthermore, the system under development will be described in more details. A particular focus of the development has been on reuse and re-configurability for task specific configurations. To that end a robotics operating systems 2 (ROS2) based system architecture, leveraging some of the improved features of ROS2 over ROS in terms of communication and microcontroller implementation through microROS (https://micro.ros.org/) has been implemented. The presentation will conclude with some examples of preliminary test and deployments.

In this work, we propose visual acoustic bundle adjustment which fuses a camera and a Doppler Velocity Log (DVL) in a graph SLAM framework for reliable underwater localization and mapping. A calibration algorithm is also designed to estimate extrinsic parameters between a camera and a DVL using features detected in scenes. Our method can achieve better robustness and localization accuracy than pure visual SLAM, especially in visually challenging underwater scenarios.