Motivation. Mobile robots usually rely on vision, ultrasonic sensing, lidar and other sensors to get information about the environment and GPS to find its location. However, there are extreme environmental conditions where those sensors do not work and where there is overall very little information available to support decision making of an autonomous robot. Therefore robots relying on unconventional sensors that can make decisions based on very little and unreliable data are necessary. Problem definition. This PhD topic involves using unconventional robotic sensors for environments where the common sensors cannot work (e.g. caves filled with slurry, ice caves, highly turbid water, etc.). The purpose will be to build world models and navigate through highly uncertain underground and underwater environments. Expected results. The expected results of this topic are physical robot prototypes which are demonstrated in real environments using a set of sensors for modelling and localising. Requirements. The successful applicant should have a M.Sc. or equivalent in Computer Science, Computer Engineering, Mechatronics or other related areas. Applicants with experience in underwater robotics or mobile robotic platforms are preferred. Apart from a strong theoretical background and practical skills in robotics and physics, this assignment requires independent thinking and creative problem solving. The work will be conducted in European Union project ROBOMINERS in an multinational team. We therefore are looking for people who are good at working with others. You will need to be prepared for field tests in unique and challenging environments, such as stone caves, ice caves and quarries underwater, including possibly unpleasant weather conditions.