RPMTD: A Route Planning Model With Consideration of Tourists’ Distribution

The rapid growth of the tourism industry has spurred extensive research into tourist route planning. However, existing studies primarily focus on route planning for individual tourists, leaving a notable gap in addressing multiple tourists planning scenarios. Traditional methods for multiple tourists planning, derived from single tourist frameworks, often prioritize the interests and benefits of tourists. This has led to problems like popularity-biased route planning, which exacerbates overtourism in favored destinations and under-tourism in lesser-known spots, thereby undermining sustainable tourism. To address these challenges, we introduce a multi-agent reinforcement learning (RL) framework specifically designed for multiple tourists route planning, incorporating consideration of tourists’ distribution. Our approach consists of two key components: firstly, a novel tourism RL environment that enables interaction with multiple tourists; secondly, a dual-congestion model that considers both local congestion at tourist spots and overall city-wide tourist’ distribution. This dual-congestion mechanism formulates the reward system in our multi-agent RL framework. We validate our framework with comprehensive experiments using real-world human mobility data from Kyoto, a globally renowned tourist city. The results show our model’s superiority over existing methods in optimizing route rewards and managing tourists’ distribution. Additionally, we conducted a user study to examine the effect of our method on tourists’ experience, serving as a reference for our future implementation. The findings indicate that our dual-congestion model slightly affects tourists who prefer popular spots. Our discussion highlights the typically non-cooperative relationship between tourism sustainability and tourist self-interest, underscoring the necessity for a trade-off between them in practical applications. Significantly, our model demonstrates the potential to transform this non-cooperative relationship into a cooperative dynamic.