Fault Recovery-Oriented Dynamic Service Migration Deployment of UGV Formations

To address the issue of vehicle functionality loss caused by processor failure in unmanned vehicle cooperative formation scenarios, this paper constructs a fault-severity-aware and scenario-adaptive migration strategy that dynamically matches the actual severity of each fault event to accommodate service migration requirements across diverse scenarios. A multi-objective evaluation model is established, incorporating migration energy cost, load balancing rate, and service migration time, to clearly define the optimization objectives for service migration strategies. We propose Whale Optimization Algorithm (WOA) to solve the multi-objective optimization problem for intra-vehicle migration and to resolve the inherent difficulty of simultaneously achieving scheduling effectiveness, rapid responsiveness, and adaptability to dynamic environments in inter-vehicle fault-induced service migration. We further develop an adaptive DQN-based dynamic service migration method with spectral clustering-based migration atomic groups (SMAG-DQN). SMAG-DQN introduces a spectral clustering-based migration atomic group constraint mechanism, integrating spectral clustering with deep reinforcement learning to construct a decision-making framework. By leveraging clustering to reduce the decision space, it effectively compresses computation time while maintaining scheduling performance. The results demonstrate that the proposed dynamic service migration strategy for UGV formations achieves significant advantages in balancing multi-dimensional performance metrics under dynamic conditions. Specifically, the WOA-based strategy attains a migration delay of 118 ms and energy cost of 8 J with 100% intra-vehicle coverage, while the SMAG-DQN-based strategy completes each decision in 0.55 s, compresses the decision chain from 78 to 15–20 steps, and achieves a delay coefficient of variation of 5.7%, demonstrating strong decision stability and effectively supporting fault recovery in cooperative UGV formation operations.