Master Thesis in Deep Reinforced Cognitive Sensing for Enhanced Automotive Radar Perception

Cognitive sensing describes the idea of a sensor system actively engaging with its surroundings in a favorable manner. This research aims to address the challenges in Automotive Radars of detecting weak targets and vulnerable road users at long distances and in variable complex environments, overcoming the limitations of current static sensor modulations. During your master thesis, you will analyze and implement a reinforced cognitive model for radar, allowing for the adaptation of the sensor modulation based on the measured environment to enhance radar perception. You will extend the existing radar simulation to establish a closed-loop pipeline for online reinforcement learning. Finally, you will train a reinforcement learning agent optimizing sensor modulation parameters under different conditions and driving scenarios. The results are to be evaluated and benchmarked against baseline methods in different use cases.