The purpose of target tracking is to represent information acquired by repeated measurement of a search volume in terms of the probabilities of objects and their states in a way that is meaningful to a system user.
Such objects commonly have models (reflecting the laws of physics) which have predictive capabilities that likelihood-wise can differentiate between correct and incorrect data-associations. However, a search volume also contains objects that are unsuited to be modeled individually. This may for example concern road-traffic, with cars only intermittently being observed, or other clutter objects that may have a broad variety of properties.
Background objects may collectively be modeled by means of an in-homogeneous Poisson process. To evaluate the likelihood for that a detection arises from the background it is necessary to be able to evaluate the intensity/probability distribution for detections from the background.
Although the constituent objects of the background are not modeled in detail, the accumulated probability distribution for detections from such objects may exhibit a fair degree of structure, being a mixture of manifolds of varying dimension.
The estimation of probability distributions defined on mixtures of manifolds is a problem that in recent years has been paid significant attention in the machine learning community.
In this thesis project, we wish to investigate the applicability of methods thereof in order to more accurately be able to represent the clutter environment perceived by our radar sensors.
Given that we need to be able to evaluate the learned distribution, we are in particular interested in methods that allow this, e.g. normalizing flows.
This provides an opportunity to possibly improve tracking performance, especially for targets in dense clutter environments (e.g. slow and small targets), which are of interest to observe and track in modern radar systems.
We are looking for one or two students at the end of master studies in either mathematics or physics. In order to successfully complete the project, you will need a strong background in mathematics and preferably knowledge of Bayesian statistics, as well as experience of machine learning.
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
Saab is a leading defence and security company with an enduring mission, to help nations keep their people and society safe.
Empowered by its 18,000 talented people, Saab constantly pushes the boundaries of technology to create a safer, more sustainable and more equitable world.
Business area Surveillance offers world-leading sensor technology in monitoring and decision support to protect against threats. The portfolio covers airborne, ground-based and naval radar, electronic warfare, C4I solutions, aviation systems and cyber security.
Affärsenheten Radar Solutions ansvarar för radarsystem för flyg, mark och sjö. Du kommer att tillhöra vår utvecklingsorganisation Engineering i Göteborg under avdelningen Signal and Data Processing Application som ansvarar för mjukvaruutveckling. Sektionen består av ca 150 personer.
Saab is a company with a strong people-orientation. We offer a friendly work environment where we support and help each other to be at our best. Continuous learning, career & talent development and employee well-being are examples of areas where we always put the strongest effort to offer great opportunities.