Towards Realistic Evaluation of Collective Perception for Connected and Automated Driving
Georg Volk, Quentin Delooz, Florian A. Schiegg, Alexander von Bernuth, Andreas Festag, Oliver Bringmann
Abstract
Collective perception in Vehicle-to-Everything (V2X) communications allows vehicles to exchange preprocessed sensor data with other traffic participants. It is currently standardized by ETSI as a second generation V2X communication service. The use of collective perception as a communication service for future fully autonomous driving requires a thorough evaluation and validation. Most of the previous work on collective perception has considered large scale-simulations with a focus on communications. However, the perception pipeline used for collective perception is equally important and must not be neglected or over-simplified. Also, to study collective perception in detail, large-scale field testing is practically infeasible. In this paper we extend an existing simulation framework with a realistic model for V2X communications and sensor-data based processing delays. The result is a simulation framework that incorporates the entire collective perception pipeline, which enables to comprehensively study sensor-based perception. We demonstrate the capabilities of this enhanced framework by analyzing the delay of each component involved in the perception pipeline. This allows a detailed insight in end-to-end delays and the age of information within the environmental model of autonomous vehicles.