Abstract:Vehicle-to-everything-aided autonomous driving (V2X-AD) has a huge potential to provide a safer driving solution. Despite extensive researches in transportation and communication to support V2X-AD, the actual utilization of these infrastructures and communication resources in enhancing driving performances remains largely unexplored. This highlights the necessity of collaborative autonomous driving: a machine learning approach that optimizes the information sharing strategy to improve the driving performance of each vehicle. This effort necessitates two key foundations: a platform capable of generating data to facilitate the training and testing of V2X-AD, and a comprehensive system that integrates full driving-related functionalities with mechanisms for information sharing. From the platform perspective, we present V2Xverse, a comprehensive simulation platform for collaborative autonomous driving. This platform provides a complete pipeline for collaborative driving. From the system perspective, we introduce CoDriving, a novel end-to-end collaborative driving system that properly integrates V2X communication over the entire autonomous pipeline, promoting driving with shared perceptual information. The core idea is a novel driving-oriented communication strategy. Leveraging this strategy, CoDriving improves driving performance while optimizing communication efficiency. We make comprehensive benchmarks with V2Xverse, analyzing both modular performance and closed-loop driving performance. Experimental results show that CoDriving: i) significantly improves the driving score by 62.49% and drastically reduces the pedestrian collision rate by 53.50% compared to the SOTA end-to-end driving method, and ii) achieves sustaining driving performance superiority over dynamic constraint communication conditions.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is how to improve driving performance by optimizing the information - sharing strategy in vehicle - to - everything communication - assisted autonomous driving (V2X - AD). Specifically, although a great deal of research has been done on traffic and communication infrastructures to support V2X - AD at present, how these infrastructures and communication resources can effectively improve driving performance in practical applications has still not been fully explored. Therefore, the paper proposes a new machine - learning method, aiming to optimize the driving performance of each vehicle through information sharing among multiple vehicles, thereby achieving collaborative autonomous driving. The main contributions of the paper include: 1. **Proposing V2Xverse**: This is a comprehensive simulation platform for collaborative autonomous driving, which can generate data to promote the training and testing of V2X - AD, and provide a complete collaborative driving process, including a multi - agent driving data set generation scheme, a code library for deploying a full - stack collaborative driving system, and a closed - loop driving performance evaluation with scene customization. 2. **Developing the CoDriving system**: This is a novel end - to - end collaborative driving system that improves driving performance by sharing perception information. CoDriving introduces a communication strategy based on driving plan feedback, allowing each intelligent vehicle to identify key driving areas and selectively request collaborative messages to enhance the perception information in these areas. 3. **Conducting comprehensive experimental verification**: Through online closed - loop evaluation, offline perception performance evaluation, and robustness evaluation, the effectiveness and superiority of the V2Xverse platform and the CoDriving system are proved. The experimental results show that, compared with the existing single - agent end - to - end driving methods, CoDriving increases the driving score by 62.49% and reduces the pedestrian collision rate by 53.50%. In conclusion, by proposing the V2Xverse platform and the CoDriving system, this paper fills the gap in the existing research on the transformation from the collaborative perception ability of multi - agent systems to the overall driving performance improvement, and provides a new solution for the development of collaborative autonomous driving.

Towards Collaborative Autonomous Driving: Simulation Platform and End-to-End System

Unified End-to-End V2X Cooperative Autonomous Driving

A Shared Control Approach for Autonomous Vehicles via Driver Behaviors Learning

End-to-End Autonomous Driving through V2X Cooperation

Collaborative Driving: Learning- Aided Joint Topology Formulation and Beamforming

A network-level connected autonomous driving evaluation platform implementing C-V2X technology

Collaborative Driving: Learning-Aided Joint Topology Formulation and Beamforming

V2X-Lead: LiDAR-based End-to-End Autonomous Driving with Vehicle-to-Everything Communication Integration

Design and Realization of Unmanned Multi‐mode Collaborative Intelligent Driving System

Cooperative Collision Avoidance Scheme Design and Analysis in V2X-Based Driving Systems

V2X-Sim: Multi-Agent Collaborative Perception Dataset and Benchmark for Autonomous Driving

Performance and Scaling of Collaborative Sensing and Networking for Automated Driving Applications

An Algorithm for Supervised Driving of Cooperative Semi-Autonomous Vehicles (Extended)

COOPERNAUT: End-to-End Driving with Cooperative Perception for Networked Vehicles

An End-to-End Collaborative Learning Approach for Connected Autonomous Vehicles in Occluded Scenarios

V2X-VLM: End-to-End V2X Cooperative Autonomous Driving Through Large Vision-Language Models

Collaborative Perception for Connected and Autonomous Driving: Challenges, Possible Solutions and Opportunities

Collaboration As a Service: Digital-Twin-Enabled Collaborative and Distributed Autonomous Driving

V2X-Communication-Aided Autonomous Driving: System Design and Experimental Validation