Intelligent route optimization for agricultural heritage tourism using deep reinforcement learning

Deep reinforcement learning significantly enhances route optimization for agricultural heritage tourism, achieving up to 31.6% improvement in sustainability compliance while increasing tourist satisfaction by 18.5% and route efficiency by 23.3% compared to traditional methods like genetic algorithms. The study introduces the Dueling Double Deep Q-Network (D3QN) framework, which integrates multi-objective constraints such as seasonal farming activities, ecological sensitivity, and socio-cultural heritage preservation into dynamic route planning for UNESCO World Heritage Sites like the Honghe Hani Rice Terraces. Operating in a simulated environment, D3QN demonstrates real-time inference capabilities with low latency (~150ms per route) and adaptability to varying tourist preferences, offering a promising solution for sustainable tourism management in agricultural heritage systems.

Sun, J. (2026). Intelligent route optimization for agricultural heritage tourism using deep reinforcement learning. Discover Artificial Intelligence. https://doi.org/10.1007/s44163-026-01895-9