Automotive Radar Detection Level Modeling with Neural Networks
Abstract: Radars are widely used in automotive to deliver both interior and exterior environment sensing capability and enable functions such as cabin monitoring, automatic emergency breaking, adaptive cruise control etc. Complex verification of automotive radars is needed to ensure the safety of the final product. Validation is performed on many levels such as climate chambers tests, electromagnetic emission compliance, algorithm robustness and much more. Computer simulations are also widely used for this purpose. Automotive radars are modelled on several levels of abstraction starting from the physical wave propagation analysis to high level system output simulation. In the article a new approach to the detection level simulation of automotive radar is presented. It is based on the neural network system modelling. The performance verification based on publicly available nuScenes dataset is presented that allows to compare the results between different models.
Loading