Efficient systems for material recognition

Purpose and goal:

The project has achieved the aim of increasing knowledge of radar and machine learning for material classification. The conducted study on system efficiency shows that an improvement of 10X lower energy consumption per classification is within reach. Expected results and effects:

This prestudy has allowed us to identify use cases and technology areas for continued research in opportunities for radar and machine learning. We have also identified parties that could be part of a subsequent project where the aim would be to have a Radar-ML material classification tool chain defined for coming sensor generations.

Approach and implementation:

This prestudy covers several activities with the aim of increasing the knowledge in how to build efficient radar systems and its application for material recognition using machine learning.

Classification models for the material recognition have been developed, tested and integrated in a prototype for evaluating robustness versus responsiveness and how it relates to key sensor figure of merits. The collaboration in the project has been good and the aim of the project was achieved.