With the growing demand for real-time computation in vehicular services, the computing power of individual vehicles is no longer sufficient to meet the low-latency execution requirements. Vehicular Edge Computing (VEC) addresses this issue by deploying computing resources at the network edge, significantly reducing service response delay. Nonetheless, the inherent high mobility of vehicles often causes them to exit the coverage zones of current edge servers, thereby undermining task continuity and execution efficiency. Although service migration serves as a key strategy to ensure service quality, frequent migrations can lead to considerable resource overhead. To address this challenge, this paper investigates a software-defined networking (SDN)-based UAV-assisted service migration mechanism for vehicular networks. UAVs are introduced as temporary edge nodes to mitigate service interruptions caused by RSU coverage blind spots or overload conditions. The Soft Actor-Critic (SAC) algorithm is employed to optimize UAV trajectory and resource allocation strategies, with the goal of minimizing task latency. Furthermore, considering UAV energy consumption, a Lyapunov-based virtual energy deficit queue is developed to ensure adherence to energy constraints while maintaining stable and efficient UAV operations. Simulation results demonstrate that the proposed approach enhances resource management and service migration efficiency in dynamic environments.