<p>Intrusion Detection Systems (IDS) are considered essential security components for the Controller Area Network (CAN) in vehicular communications. However, existing methods struggle to adapt to varied attack scenarios and accurately detect low-volume attacks. In this paper, we introduce Meta-IDS, a novel IDS that employs meta-learning via the Meta-SGD algorithm to enhance adaptability across a diverse spectrum of cyber threats. Our method includes a bi-level optimization technique: the inner level optimizes detection accuracy for specific attack scenarios, while the outer level adjusts meta-parameters to ensure generalizability across different scenarios. To model low-volume attacks, we devise the Attack Prominence Score (APS), which identifies subtle attack patterns. Extensive experimental results show that the proposed method exhibits promising detection performance, with efficient tuning and rapid adaptation to different attack scenarios, especially low-volume attacks. Furthermore, Meta-IDS demonstrates competitive scalability in intelligent transportation systems, supporting larger network infrastructures and high throughput. Real-time vehicle-level evaluations also show that it is lightweight for vehicular networks.</p>