State-of-the-Art Sensor Models for Virtual Testing of Advanced Driver Assistance Systems/Autonomous Driving Functions

Birgit Schlager,Stefan Muckenhuber,Simon Schmidt,Hannes Holzer,Relindis Rott,Franz Michael Maier,Kmeid Saad,Martin Kirchengast,Georg Stettinger,Daniel Watzenig,Jonas Ruebsam

DOI: https://doi.org/10.4271/12-03-03-0018

2020-10-29

SAE International Journal of Connected and Automated Vehicles

Abstract:Sensor models are essential for virtual testing of Advanced Driver Assistance Systems/Autonomous Driving (ADAS/AD) functions. This article gives an overview of the state-of-the-art of ADAS/AD sensor models. The considered sensors are radar, lidar, and camera. To get a common understanding and a common language in sensor model research, a new classification method into low-, medium-, and high-fidelity sensor models is introduced. Low-fidelity sensor models are based on geometrical aspects like the Field Of View (FOV) of the sensor and object positions in the virtual environment. Object lists are used as input and output data formats. Medium-fidelity sensor models consider the detection probability and physical aspects in addition to geometrical aspects of the sensor. They have object lists as input and object lists or raw data as output. High-fidelity sensor models are based on rendering techniques. They have the virtual three-dimensional (3D) environment provided by the environment simulation as an input and sensor raw data as an output. The classification is useful for virtual testing of ADAS/AD functions since the classes can be correlated to the phases of the Systems Development Process (SDP) of ADAS/AD.