Abstract Under-ice exploration in the Arctic Ocean requires robust and reliable robotic tools. Autonomous underwater vehicles (AUVs) used in this environment have to be able to navigate reliably and with precision without the help of a dedicated infrastructure. They also have to be able to adapt and to react to changes in their environment. In this paper, new methods for AUV infrastructure-independent navigation and self-localization, as well as for an adaptive AUV control architecture are presented. These methods were developed at the DFKI within the framework of several nationally funded research projects. They were implemented on experimental AUVs and showed promising results in preliminary lab and field tests. Introduction Robots are needed to operate in areas that are too difficult or to too dangerous to access for humans. In remote areas like the deep sea and the Arctic Ocean, robots are invaluable helpers for scientific exploration, environmental monitoring, economic exploitation and other endeavours. As of today, robots have been used successfully by oceanographic research institutes, oil companies, and other stakeholders to access and explore the depths of the Arctic Ocean. However, the application of these robots - like the application of most professional service robots in harsh environments - is still hampered by a lack of autonomy, robustness and dependability. So far, robots can only be used in well-supervised short-term missions that require intensive support by an associated research vessel. A robust long-term deployment of autonomous robots in harsh environments, with a minimum of human supervision and support, is well beyond the current-state-of-the-art. In this paper, we will present new methods that lead towards solutions for two key obstacles in the way of robust operation of robots in arctic exploration, namely the problem of reliable under water and under-ice self-localization and navigation, and the problem of adaptive mission control for AUVs. Both methods are the results of ongoing projects at the Robotics Innovation Center of the German Research Center for Artificial Intelligence (DFKI-RIC) in Bremen, Germany. AUV Navigation and Self-Localization Introduction The ability to follow a given course without deviation and to determine it's own position without error is a challenge for any underwater vehicle, and in particular for AUVs that operate under ice. Nevertheless, depending on the mission profile as well as on the precision and type of self-localization required, there are numerous methods to determine the current position of an AUV.