Study with us!
One of the cornerstones of our business model is the collaboration with academics. An important part of this is student collaboration. We are always looking for students who are eager to develop the future of mobility systems with us. Have a look below on our project proposals or get inspired by what students have worked on together with us in the past.
Proposed student projects
Here’s a list of our student work proposals. Send an email if you would like to work with any of the projects with us. If you can’t find a proposal which suits you please don’t hesitate to contact us. We are happy to discuss a scope for a thesis within our business which suits your field and interests.
Robust and efficient algorithms for motion planning and collision avoidance are key to enable autonomous ferries. Moving to more complex operational domains including e.g. long transits and complex traffic situations, further research on motion planning and collision avoidance is required. Important keywords are the "rules of the road" (COLREGs), traffic interactions and online risk management.
A land-based sensor may improve the situational awareness of the ferry beyond line of sight (e.g. beyond turns and bridges). It may also be possible to move sensors from the ferry to land without reducing situational awareness. These sensors will detect the ferry in addition to other objects in the vicinity.
Research papers from the autonomous driving community have argued that stereo vision can reach sufficient accuracy to replace lidars. The student will followup work on stereo vision for milliAmpere 1 and transition it to mA2. The main goal is to investigate the performance gain of stereo vision versus lidars.
The research vessel milliAmpere has a sensor rig that contains both infrared and RGB cameras. The cameras are stacked vertically, which means that the cameras have a large overlap in the field of view. This means that objects detected in the RGB frame can be transferred into the IR camera image. This way, the relatively weaker detection methods used in IR images can be trained using this automatically labeled data.
Doppler information can be very beneficial in sensor data processing (object segmentation). For sensor fusion, the uncertain doppler measurement is weighted against more precise measurements from e.g. lidar.
The thesis will employ, adapt and extend a method called adaptive stress testing (AST) to find likely failure events of the autonomy system, using reinforcement learning (RL) and other AI and ML techniques. The AST method has previously been used to validate aircraft collision avoidance systems, which is a similar problem.
The assignment can be divided in three parallel challenges: modeling of ship wakes, modeling of how these propagate and interact with ocean waves, and how to simulate this in real time with sufficient accuracy on a GPU.
NTNU and Zeabuz have developed a Digital Twin setup for the pilot ferry milliAmpere 2, with simulation models established using the Open Simulation Platform (OSP). This model should be further developed, tuned towards the recently launched real milliAmpere 2 ferry, and updated to facilitate updates based on data streams from the ferry.
Utne et al. (2020) har proposed a method for developing an online risk model using Systems Theoretic Process Analysis (STPA) and Bayesian Belief Networks (BBNs). The student should develop an ORM system using this methodology for the milliAmpere 2 autonomous passenger ferry.
There has not been any application of FV to autonomous marine vessels yet. Pioneering work is required to explore possible uses. The thesis will represent a first-take on formal verification of autonomous marine systems.
To enable fast simulations of motion planning systems, the thesis should investigate how to emulate an object detection and situational awareness system to provide realistic input to the motion planning system.
Completed and ongoing student projects
Here you can find the finished and ongoing projects which have been carried out by students in collaboration with us so far.