Collective perception plays a critical role in enhancing the awareness and safety of Intelligent Transportation Systems (ITS). However, most existing studies on collective perception have relied on ideal perception models, neglecting the challenges and limitations of real-world perception. To address this research gap, this paper presents the implementation of a comprehensive simulation framework specifically designed to incorporate realistic perception in the Collective Perception Service. To achieve realistic perception, a comprehensive perception pipeline is developed, integrating OpenCDA [1] and state-of-the-art 3D object detection algorithms [2] named PointPillars. Artery [3] serves as the communication component, facilitating V2X communication within the framework. Additionally, a middleware is designed to establish seamless integration between perception and communication components. By comparing the results of realistic perception against ideal perception, we analyze the impact of perception modeling on the performance of collective perception systems. The findings indicate that the realistic perception model achieves approximately 42% of the detection capability of the ideal perception model. Moreover, the evaluation demonstrates that realistic perception leads to a lower channel busy ratio while maintaining an acceptable level of environmental awareness, when compared to ideal perception.
A Study on Collective Perception with Realistic Perception Modeling
2023-10-10
6315878 byte
Conference paper
Electronic Resource
English
TIBKAT | 2022
|Impact of Perspectives on Collective Perception
IEEE | 2024
|Realizing collective perception in a vehicle
IEEE | 2016
|