The increase in tourists in a destination has exceeded the capacity of the public transportation system, creating traffic congestion. It has led to lower levels of tourist satisfaction. We proposes a reinforcement learning-based method for planning tourist routes, which accounts for congestion not only at popular tourism spots but also in transportation and road conditions. By implementing dynamic reward functions and constraints, our method demonstrates through experimental results a reduction in congestion levels at spots and a decrease in route congestion between these spots compared to existing approaches.