Bi-LSTM based Multi-Agent DRL with Computation-aware Pruning for Agent Twins Migration in Vehicular Embodied AI Networks

2025-05-09

AI Summary

To address high inference latency and poor service continuity in onboard embodied AI agents for connected and autonomous vehicles (CAVs) caused by onboard computational constraints, this paper proposes a dynamic digital twin migration mechanism orchestrated by roadside units (RSUs). We introduce the first AV-RSU Stackelberg game model to formalize vehicle–infrastructure cooperative decision-making. A lightweight, decentralized multi-agent bidirectional LSTM-PPO algorithm is designed to achieve equilibrium solutions without centralized coordination. Furthermore, we propose Path eXclusion (PX), a personalized neural network pruning method tailored to heterogeneous onboard terminals. Experimental results demonstrate that our approach reduces average migration latency by 38.7%, decreases RSU load variance by 52.4%, and achieves 4.3× model compression while preserving 98.2% model accuracy—significantly enhancing edge collaboration efficiency and real-time performance.

Technology Category

• Multiagent Systems: Mechanism Design
• Machine Learning: Learning on the Edge & Model Compression
• Planning, Routing, and Scheduling: Model-Based Reasoning

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

With the advancement of large language models and embodied Artificial Intelligence (AI) in the intelligent transportation scenarios, the combination of them in intelligent transportation spawns the Vehicular Embodied AI Network (VEANs). In VEANs, Autonomous Vehicles (AVs) are typical agents whose local advanced AI applications are defined as vehicular embodied AI agents, enabling capabilities such as environment perception and multi-agent collaboration. Due to computation latency and resource constraints, the local AI applications and services running on vehicular embodied AI agents need to be migrated, and subsequently referred to as vehicular embodied AI agent twins, which drive the advancement of vehicular embodied AI networks to offload intensive tasks to Roadside Units (RSUs), mitigating latency problems while maintaining service quality. Recognizing workload imbalance among RSUs in traditional approaches, we model AV-RSU interactions as a Stackelberg game to optimize bandwidth resource allocation for efficient migration. A Tiny Multi-Agent Bidirectional LSTM Proximal Policy Optimization (TMABLPPO) algorithm is designed to approximate the Stackelberg equilibrium through decentralized coordination. Furthermore, a personalized neural network pruning algorithm based on Path eXclusion (PX) dynamically adapts to heterogeneous AV computation capabilities by identifying task-critical parameters in trained models, reducing model complexity with less performance degradation. Experimental validation confirms the algorithm's effectiveness in balancing system load and minimizing delays, demonstrating significant improvements in vehicular embodied AI agent deployment.

Problem

• Optimize bandwidth allocation for AI agent migration in VEANs
• Balance workload among RSUs using Stackelberg game modeling
• Reduce model complexity via computation-aware pruning for AVs

Innovation

• Bi-LSTM based Multi-Agent DRL for agent twins migration
• Stackelberg game optimizes AV-RSU bandwidth allocation
• Personalized PX pruning adapts to AV computation capabilities