Thèse CIFRE : Méthodes géométriques dans les radars automobiles F/M

Context:

NXP is a leader in the microprocessors used in automotive radar sensors. As the demand of new products and solutions to difficult road conditions increases, new radar processing algorithms and paradigms are sought by our customers.

The theory of radar measurements can be considered from a geometric viewpoint. Problems in deterministic and random signal detection have a modern foundation in information geometry. In such a framework, probability distributions are seen as geometrical objects on the statistical manifold. Data with values in the space of 𝑚 by 𝑚 Hermitian positive definite matrices, plays an essential role in many applications, including medical imaging, computer vision and radar signal processing.

This project seeks to find application of these techniques in the automotive radar space. In particular, the improvement of the detection and perception limitation capabilities of ADAS systems. These might include NCAP scenarios such as pedestrian detection, debris and pothole awareness, recognition of tunnels and bridges and sensitivity improvement of automotive sensors of hard to detect objects and situations.

Duration : Three-year Ph.D. project

Your responsibilities

The Objectives are to

• Understand the fundamentals of automotive radar physics.

• Acquire familiarity with Lie group representation theory.

• Utilize existing toolboxes for statistical learning in manifolds such as Geomstats: A Python Package for Riemannian Geometry

• Algorithmic Development with mathematical models and simulations.

• Performance evaluation and/or implementation in Hardware.

Your profile

• Master thesis level in Physics, Mathematics or Engineering with an interest in algorithm development.

• Linear algebra beyond the first course.

• Knowledge of Probability theory and random variables

• An advanced mathematics course such as Group Theory and/or Differential Geometry.

• Good programming skills is required (Python or C is a plus).

• Knowledge of time frequency analysis, embedding programming and/or electronic circuits is a plus.

More information about NXP in France...