Thesis Work: Motion Planning within the Latent Space of Large Maps (30HP)
Saab is continously developing and improving our technical edge over competition, hence we are very interested to exploit the possibilities that machine learning can bring us in the field of motion planning of airborne products.
Description of master thesis
A problem within motion planning is the scalability of methods to large environments. Hence, this thesis aims to investigate motion planning within the compressed representation of large maps, and to do so using machine learning-based techniques. In essence, there are many aspects to this problem among which you may choose a sub-problem to focus your thesis on. For example, what is an appropriate way of down-sampling a large map or costmap? How should the planning algorithm be designed? Of what form should the training data be if you opt for a supervised-oriented solution? If instead you choose to develop a reinforcement-based algorithm, how will you devise it?
The actual project scope will be adapted to appropriate content in dialogue with the student.
This project is suitable for you that will conduct your master thesis within the engineering field of automation, simulation, electrical or data engineering. Preferably you have experience within programming, linear algebra and a familiarity with reading scientiﬁc articles. Prior knowledge about machine learning is preferred but not required.
The work shall preferably be conducted at SAAB Dynamics in Karlskoga.
Peter Andersson, Chef
If you aspire to help create and innovate whilst developing yourself in a challenging team setting, Saab may well have the perfect conditions for you to grow. We pride ourselves on a nurturing environment, where everyone is different yet we share the same goal – to help protect people.