This paper presents a didactic proof-of-concept for a hybrid quantum–classical framework for optimizing drone activation and coordination using Grover’s quantum search algorithm combined with a neural network trained on quantum-sampled data. The proposed approach models the activation of each drone as a qubit, encoding feasible mission configurations in a high-dimensional binary state space. A Grover-based oracle is constructed to mark energetically favorable configurations that balance mission coverage, battery constraints, and inter-drone interference. Due to current simulation constraints, the problem is evaluated at a small scale ( $n=10$ ) to focus on the pipeline’s mechanics rather than demonstrating immediate quantum supremacy. The resulting quantum circuit, implemented in Cirq, is simulated to obtain probabilistic samples of low-energy states, which are then used to train a feed-forward neural network. The trained model successfully generalizes the energy landscape and predicts optimal configurations without exhaustive enumeration. Numerical experiments confirm that quantum-sampled training data enhance neural prediction accuracy and efficiently approximate the global optimum found by classical brute-force search. This study demonstrates how Grover’s algorithm can serve as a quantum data generator to improve classical learning models, serving as an exploratory step toward future, larger-scale implementations.
Optimization of vehicle-to-station assignments under capacity and distance constraints represents a challenging combinatorial problem relevant to automated planning, logistics, and autonomous mobility systems. Classical methods such as Mixed Integer Linear Programming (MILP) or metaheuristics often struggle to scale efficiently with problem dimensionality, motivating the exploration of hybrid quantum-classical paradigms. This paper presents a normalized Quantum Approximate Optimization Algorithm (QAOA) framework tailored for constrained assignment problems, where vehicle-station distances are encoded into a normalized cost Hamiltonian. Capacity violations and unused resources are incorporated through dynamically scaled penalty terms, producing a cost landscape that effectively guides the quantum search process. The proposed pipeline integrates parameter optimization using COBYLA to refine the QAOA angles, ensuring convergence toward low-cost feasible configurations. Experimental simulations in Cirq on a 10-qubit system demonstrate that the normalized QAOA pipeline consistently identifies near-optimal assignments while substantially reducing the combinatorial search space. These results provide empirical evidence for the viability of hybrid QAOA formulations in real-world planning and scheduling scenarios, establishing a foundation for future implementations on noisy intermediatescale quantum (NISQ) hardware.
Jakub Kołota· International Conference on...· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
