Quantum Machine Learning for Intelligent Transportation: A Survey

Intelligent Transportation Systems (ITS) face escalating demands for ultra-low latency, quantum-resilient security, efficient processing of heterogeneous high-volume data, and strict resource constraints—pressures that increasingly exceed the capabilities of classical architectures. Quantum Machine Learning (QML) and quantum-enhanced security provide a transformative pathway by enabling quantum-accelerated combinatorial optimization, compact variational inference, and information-theoretic cryptography. This survey delivers a deployment-oriented synthesis of QML for ITS, proposing a multi-layer hybrid quantum–classical architecture that maps core ITS functions—collision avoidance, traffic prediction, perception, and secure V2X communications—to appropriate quantum substrates and security primitives (post-quantum cryptography, quantum homomorphic encryption, quantum differential privacy) across edge, RSU, and cloud tiers. We introduce an operational taxonomy linking ITS challenges to quantum computational archetypes, formalize algorithmic foundations using resource-aware complexity models (qubits, depth, shots), and provide a unified comparison of quantum versus classical pipelines using harmonized metrics (latency, accuracy, energy, cost). By integrating algorithmic rigor, empirical evidence, deployment feasibility, and security co-design, this work establishes a prescriptive foundation for quantum-enhanced ITS in both the NISQ and early fault-tolerant eras. Specifically, quantum annealing supports large scale ITS optimization by casting tasks such as traffic coordination, routing, and V2X assignment as QUBO/Ising models whose low-energy states correspond to near-optimal solutions, while quantum reinforcement learning combines a classical actor with a quantum critic—implemented via parametrized quantum circuits or Quantum Boltzmann Machines—to achieve more expressive value estimation and superior sample efficiency for sequential ITS decision-making.