Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

PACP: Priority-Aware Collaborative Perception for Connected and Autonomous Vehicles

Zhengru Fang, Senkang Hu, Haonan An, Yuang Zhang, Jingjing Wang, Hangcheng Cao, Xianhao Chen, Yuguang Fang

TL;DR

This paper tackles BEV blind spots in connected autonomous vehicle perception by introducing PACP, a priority-aware collaborative perception framework that uses a BEV-match mechanism to prioritize data from nearby CAVs. It couples a two-stage optimization (nonlinear programming and submodular optimization) with a deep learning–driven adaptive autoencoder to maximize a weighted utility that accounts for perception quality and coverage under realistic V2V constraints. The approach is validated on CARLA/OpenCOOD with the OPV2V dataset, showing improvements in utility and AP@IoU over state-of-the-art baselines (e.g., up to 8.27% utility gain and 13.60% improvement in AP@IoU). The combination of IoU-based priority weighting, submodular optimization, and RSU-enabled fine-tuning yields robust performance under dynamic channel conditions, highlighting practical potential for real-time, high-assurance cooperative perception in urban CAV networks.

Abstract

Surrounding perceptions are quintessential for safe driving for connected and autonomous vehicles (CAVs), where the Bird's Eye View has been employed to accurately capture spatial relationships among vehicles. However, severe inherent limitations of BEV, like blind spots, have been identified. Collaborative perception has emerged as an effective solution to overcoming these limitations through data fusion from multiple views of surrounding vehicles. While most existing collaborative perception strategies adopt a fully connected graph predicated on fairness in transmissions, they often neglect the varying importance of individual vehicles due to channel variations and perception redundancy. To address these challenges, we propose a novel Priority-Aware Collaborative Perception (PACP) framework to employ a BEV-match mechanism to determine the priority levels based on the correlation between nearby CAVs and the ego vehicle for perception. By leveraging submodular optimization, we find near-optimal transmission rates, link connectivity, and compression metrics. Moreover, we deploy a deep learning-based adaptive autoencoder to modulate the image reconstruction quality under dynamic channel conditions. Finally, we conduct extensive studies and demonstrate that our scheme significantly outperforms the state-of-the-art schemes by 8.27\% and 13.60\%, respectively, in terms of utility and precision of the Intersection over Union.

PACP: Priority-Aware Collaborative Perception for Connected and Autonomous Vehicles

TL;DR

This paper tackles BEV blind spots in connected autonomous vehicle perception by introducing PACP, a priority-aware collaborative perception framework that uses a BEV-match mechanism to prioritize data from nearby CAVs. It couples a two-stage optimization (nonlinear programming and submodular optimization) with a deep learning–driven adaptive autoencoder to maximize a weighted utility that accounts for perception quality and coverage under realistic V2V constraints. The approach is validated on CARLA/OpenCOOD with the OPV2V dataset, showing improvements in utility and AP@IoU over state-of-the-art baselines (e.g., up to 8.27% utility gain and 13.60% improvement in AP@IoU). The combination of IoU-based priority weighting, submodular optimization, and RSU-enabled fine-tuning yields robust performance under dynamic channel conditions, highlighting practical potential for real-time, high-assurance cooperative perception in urban CAV networks.

Abstract

Surrounding perceptions are quintessential for safe driving for connected and autonomous vehicles (CAVs), where the Bird's Eye View has been employed to accurately capture spatial relationships among vehicles. However, severe inherent limitations of BEV, like blind spots, have been identified. Collaborative perception has emerged as an effective solution to overcoming these limitations through data fusion from multiple views of surrounding vehicles. While most existing collaborative perception strategies adopt a fully connected graph predicated on fairness in transmissions, they often neglect the varying importance of individual vehicles due to channel variations and perception redundancy. To address these challenges, we propose a novel Priority-Aware Collaborative Perception (PACP) framework to employ a BEV-match mechanism to determine the priority levels based on the correlation between nearby CAVs and the ego vehicle for perception. By leveraging submodular optimization, we find near-optimal transmission rates, link connectivity, and compression metrics. Moreover, we deploy a deep learning-based adaptive autoencoder to modulate the image reconstruction quality under dynamic channel conditions. Finally, we conduct extensive studies and demonstrate that our scheme significantly outperforms the state-of-the-art schemes by 8.27\% and 13.60\%, respectively, in terms of utility and precision of the Intersection over Union.
Paper Structure (33 sections, 3 theorems, 35 equations, 15 figures, 4 tables, 2 algorithms)

This paper contains 33 sections, 3 theorems, 35 equations, 15 figures, 4 tables, 2 algorithms.

Table of Contents

  1. Introduction
  2. Background
  3. State-of-the-Art
  4. Our Contributions
  5. Fairness or Priority?
  6. Background of Two Schemes
  7. An Illustrative Motivating Example
  8. System Model
  9. System Overview
  10. Channel Modeling and System Constraints
  11. Priority-Aware Collaborative Perception Architecture
  12. Selection of Priority Weights
  13. Procedure of Obtaining Priority Weights
  14. Utilization Maximization in Collaborative Perception
  15. Two-Stage Optimization Framework for Utility Maximization
  16. ...and 18 more sections

Key Result

Proposition 1

For fixed $\mathcal{D}^{(n)}$, problem $\mathbf{P}$ is reducible from Weighted Maximum Coverage Problem (WMCP). When optimizing over matrices $\mathcal{D}$ and $\mathcal{R}$, $\mathbf{P}$ surpasses WMCP's complexity, establishing its NP-hardness.

Figures (15)

  • Figure 1: The bandwidth and throughput allocation by different schemes within V2V network.
  • Figure 2: Camera data and different types of AP@IoU.
  • Figure 3: Overview of V2X-aided collaborative perception system.
  • Figure 4: Procedure for priority weight calculation. Fig. \ref{['fig:pacs']}(a): CAVs observe surroundings with 4 cameras; CAVs 1-2 relay RGB data to CAV 0. Figs. \ref{['fig:pacs']}(b)-(c): BEV feature generation in CAV 0. Fig. \ref{['fig:pacs']}(d): BEV-match mechanism determines priority weights.
  • Figure 5: An example of determining priority weight $\mathcal{P}_{10}$.
  • ...and 10 more figures

Theorems & Definitions (11)

  • Proposition 1
  • Proof 1
  • Definition 1
  • Definition 2
  • Definition 3
  • Proposition 2
  • Proof 1
  • Proposition 3
  • Proof 2
  • Proof 3
  • ...and 1 more