Ships navigate within maritime networks, which have evolved from extensive accumulated experience over time. Therefore, maritime networks generated from Automatic Identification System (AIS) data can effectively guide ship navigation. This study aims to address humanoid decision-making for autonomous vessel navigation systems, proposing a method for generating habitual maritime route networks based on trajectory data spatiotemporal feature mining and machine learning techniques. This paper first conducts spatiotemporal feature analysis of vessel behaviors using Automatic Identification System (AIS) data to identify crucial route points such as stopping points and Waypoint. Subsequently, DBSCAN clustering and Douglas-Peucker algorithm were used to process the navigation points and extract the key turning nodes of the habitual route network. The research findings indicate that the generated habitual maritime route network effectively reflects the main routes and node distribution of maritime traffic, identifying 3683 docking node regions and generating 3000 turning node regions, forming a clear maritime traffic network. Finally, a coherent chain is formed which is modeled as a directed graph. The proposed method can effectively realize the construction of habitual route networks.