In this thesis, we aim to investigate deep learning architectures that efficiently capture the relevant information for a decision-making agent trained by reinforcement learning for beyond visual range air combat.
Your role
Background
Reinforcement Learning (RL) offers transformative potential for future fighter jets by enabling autonomous or human-assisting decision-making and adaptive behaviours in complex, dynamic environments. In a typical mission, the pilot or a decision-making agent in the aircraft needs to make sense of an ever-increasing number of dynamic objects that may or may not be relevant to the overall mission goals. We therefore study both decision-making algorithms as well as agent representations, to see which may be effective in processing, predicting and supporting decisions based on complex data, but also for learning in dynamic scenarios.
Description of the master thesis
This Master Thesis aims at studying deep learning representations that may be relevant for building decision-making agents, by training them in simulated beyond visual range air combat, with a focus on spatial-relational and temporal concepts. This could include evaluating spatial processing architectures such as 360-degree convolutions, vision transformers and graph neural networks, or temporal processing architectures such as transformers, state space models and recurrent networks. The exact specifics will be decided by you and our supervisors so that your input can help shape how we focus our investigation.
You will be part of a research-intensive unit called AI-powered mission autonomy that include about 20 people including PhDs, PhD students and Research Engineers focusing on reinforcement learning, optimization, planning, and AI operations. The unit is involved in projects from early research all the way to product development and support with expert knowledge and the development of AI-based functions.
Your profile
This Master Thesis is suitable for 1-2 students with interest in Deep Reinforcement Learning as well as Deep Learning. Note that training complex architectures in Deep Reinforcement Learning may be non-trivial compared to training them in a supervised learning setting.
You are at the end of your Master of Science in e.g., Computer Science and Engineering or Engineering Physics and about to start your Master Thesis work for 30 HP.
This position requires that you pass a security vetting based on the current regulations around/of security protection. For positions requiring security clearance additional obligations on citizenship may apply.
What you will be a part of
Behind our innovations stand the people who make them possible. Brave pioneers and curious minds. Everyday heroes and inventive troubleshooters. Those who share deep knowledge and those who explore sky-high. And everyone in between.
Joining us means making an impact together, contributing in our own unique ways. From crafting complex code and building impressive defence and security solutions to simply sharing a coffee with a colleague, every action counts. We encourage you to take on challenges, to create smart inventions and grow in our friendly and tech-savvy workspace. We have a solid mission to keep people and society safe.
Saab is a leading defence and security company with an enduring mission, to help nations keep their people and society safe. Empowered by its 23,000 talented people, Saab constantly pushes the boundaries of technology to create a safer and more sustainable world. Saab designs, manufactures and maintains advanced systems in aeronautics, weapons, command and control, sensors and underwater systems. Saab is headquartered in Sweden. It has major operations all over the world and is part of the domestic defence capability of several nations. Read more about us here
Contact information
Emil Karlsson, Manager
0734-180368
Dennis Malmgren, Master Thesis Supervisor
0734-181391
dennis.malmgren1@saabgroup.com
Kindly observe that this is an ongoing recruitment process and that the position might be filled before the closing date of the advertisement