Deployment of deep neural networks onto embedded devices for processing radar data
Recently Artificial Intelligence (AI) and Machine Learning techniques are used in a wide range of applications, for example in autonomous driving. It allows to perform real-time obstacle and object detection based on radar, camera and lidar data. Processing radar data on-board is a demanding task due to high bandwidth of input data combined with limited computational resources.
In this thesis the application of deep neural networks onto embedded devices for processing radar data will be evaluated. This approach will be used in an existing vehicle system. In order to meet the requirements of the system, optimizations such as quantization, pruning will be investigated. The optimized neural network will be evaluated on different microcontroller-based hardware platforms with respect to latency and resource overhead to meet the requirements of the system, such as energy efficiency, performance and latency.
Programming experience in Python and C/C++;
Basic knowledge of neural networks;
Basic knowledge of microcontrollers;
Motivation and interest in solving technical problems independently.