Optimizing Autonomous Vehicle Sensor Setups: A Framework for Coverage Analysis

The performance of environment perception in autonomous vehicles is significantly influenced by the sensor setup, which is determined in the early design phase. The selection of the type and position of sensors in this phase often occurs before the availability of data processing algorithms. Therefore, the evaluation of the sensor setup can only be based on coverage metrics. This paper presents a novel framework for modeling the sensor coverage of autonomous vehicles. By discretizing the environment into grid cells, our framework analyzes each sensor and the whole setup. Our methodology systematically determines the coverage, identifies redundancy and blind spot areas, and obtains quantifiable metrics for evaluating the sensor setup efficiency. Utilizing the PyVista visualization library, we present the individual building blocks of the framework and their open-source implementation. The results from a real-world case study demonstrate our framework's ability to identify weaknesses in sensor setup coverage. Our approach helps to develop a comprehensive understanding of the sensor coverage and, therefore, contributes to designing more effective and reliable sensing systems.