With the development of the social economy and the improvement of living standards, the number of privately owned vehicles is sustainable increasing, so intelligent parking system is a research direction currently. The manual management and guidance of traditional parking lots has many issues such as low efficiency and poor customer experience. To address these issues, this paper designs an intelligent parking lot system that amalgamates data collection at the site, cloud processing, digital human guidance and interaction. In terms of path planning, it optimizes the algorithm based on reinforcement learning which using a dynamic greedy factor, and further refining existing path planning by reducing the probability of vehicle encounters. The system fills the gap in the application of AI digital human within the field of transportation, and the digital human avatars can provide more friendly communication for users. Meanwhile, it achieves better interactive effects and provides a reliable solution for improving the efficiency of parking lot management.