Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Cloud-Edge-Terminal Collaborative AIGC for Autonomous Driving

Jianan Zhang, Zhiwei Wei, Boxun Liu, Xiayi Wang, Yong Yu, Rongqing Zhang

TL;DR

The paper addresses the latency and resource bottlenecks of deploying large AIGC models in autonomous driving by proposing a cloud-edge-terminal collaborative architecture that distributes generation and inference across centralized clouds, edge units, and vehicle terminals. It surveys AIGC applications for perception, motion planning, traffic simulation, data generation, and human–vehicle interfaces, and outlines a workflow for offloading, post-processing, and personalization. It introduces concrete mechanisms for resource management, including elastic task generation, proactive caching, and an interactive MARL (POSG) framework for ASP selection guided by a learnable Reward Model. The approach aims to enable low-latency, region-specific, and personalized AIGC-assisted driving while informing network design through integrated AI and communications strategies.

Abstract

In dynamic autonomous driving environment, Artificial Intelligence-Generated Content (AIGC) technology can supplement vehicle perception and decision making by leveraging models' generative and predictive capabilities, and has the potential to enhance motion planning, trajectory prediction and traffic simulation. This article proposes a cloud-edge-terminal collaborative architecture to support AIGC for autonomous driving. By delving into the unique properties of AIGC services, this article initiates the attempts to construct mutually supportive AIGC and network systems for autonomous driving, including communication, storage and computation resource allocation schemes to support AIGC services, and leveraging AIGC to assist system design and resource management.

Cloud-Edge-Terminal Collaborative AIGC for Autonomous Driving

TL;DR

The paper addresses the latency and resource bottlenecks of deploying large AIGC models in autonomous driving by proposing a cloud-edge-terminal collaborative architecture that distributes generation and inference across centralized clouds, edge units, and vehicle terminals. It surveys AIGC applications for perception, motion planning, traffic simulation, data generation, and human–vehicle interfaces, and outlines a workflow for offloading, post-processing, and personalization. It introduces concrete mechanisms for resource management, including elastic task generation, proactive caching, and an interactive MARL (POSG) framework for ASP selection guided by a learnable Reward Model. The approach aims to enable low-latency, region-specific, and personalized AIGC-assisted driving while informing network design through integrated AI and communications strategies.

Abstract

In dynamic autonomous driving environment, Artificial Intelligence-Generated Content (AIGC) technology can supplement vehicle perception and decision making by leveraging models' generative and predictive capabilities, and has the potential to enhance motion planning, trajectory prediction and traffic simulation. This article proposes a cloud-edge-terminal collaborative architecture to support AIGC for autonomous driving. By delving into the unique properties of AIGC services, this article initiates the attempts to construct mutually supportive AIGC and network systems for autonomous driving, including communication, storage and computation resource allocation schemes to support AIGC services, and leveraging AIGC to assist system design and resource management.
Paper Structure (9 sections, 5 figures)

This paper contains 9 sections, 5 figures.

Table of Contents

  1. Introduction
  2. AIGC for autonomous driving
  3. AIGC Applications to Autonomous Driving
  4. Challenges of Applying AIGC to Autonomous Driving
  5. Cloud-Edge-Terminal Collaborative AIGC Architecture
  6. Architecture Design
  7. System Operations
  8. Conclusion
  9. Acknowledgments

Figures (5)

  • Figure 1: Applications of AIGC to autonomous driving.
  • Figure 2: The cloud-edge-terminal collaborative AIGC architecture.
  • Figure 3: AIGC for elastic task generation, communication resource prediction and allocation.
  • Figure 4: Proactive caching based on AIGC estimation and prediction for autonomous driving.
  • Figure 5: The interactive MARL framework for ASP selection based on drivers' personalized reward models.