Sensor Modeling for Close Range Automotive Environmental Perception Systems

Abstract

This doctoral project aims to bridge the gap between simulation and real-world autonomous driving by developing accurate models for wide Operational Design Domains. The research focuses on modeling raw, high-resolution ultrasonic signals and camera data under realistic adverse conditions like rain, varying ground textures, and complex lighting. By employing a hybrid approach that integrates physical laws with machine learning, the project ensures that generated data remains physically consistent. Additionally, the study investigates pioneering cross-modal signal prediction, such as using camera images to compute corresponding ultrasonic measurements. The aim is to accelerate the development of safer autonomous technology by reducing the industry's reliance on costly physical road testing.