Modeling and Analysis of Collaborative Vision-Based Environment Perception in Vehicular Networks With Visual Blockage

Collaborative vision-based environment perception is envisioned as a critical component of autonomous driving in intelligent transportation systems. In collaborative vision-based environment perception, vehicular nodes interact and transfer their local visual information to overcome occlusions induced by surrounding obstacles. Despite significant algorithmic advances in collaborative vision-based environment perception, the development of a comprehensive probabilistic framework to study its performance remains insufficiently explored. Inspired by this, this paper develops a novel analytical framework for collaborative vision-based environment perception in vehicular networks by applying tools from stochastic geometry. In particular, we model the road system as a Poisson line process (PLP). Further, by leveraging concepts from random shape theory, the building blockages and vehicle blockages are modeled as independent random rectangle processes, while the irregularly shaped blockages are characterized as stochastic polygonal blockages based on Voronoi tessellations. For this setup, we derive analytical expressions for the single-vehicle perception probability, the coverage probability, and the mean area fraction of the typical vehicle. We perform Monte-Carlo simulations to validate the theoretical framework and provide system-design insights through numerical analysis. To the best of the authors' knowledge, this is the first work to analyze the system-level performance of collaborative vision-based environment perception in vehicular networks with random visual blockages.