With rapid urbanization，traditional fixed-route bus systems increasingly show limitations in service efficiency and flexibility.The Demand Responsive Transit （DRT）system，as an innovative public transportation mode， demonstrates great potential in improving service quality and operational efficiency by dynamically adjusting routes in response to real-time passenger demands.This study proposes a DRT scheduling method based on the Hybrid-Perception Adaptive Genetic Algorithm （HPAGA）.The method incorporates a hybrid-weight chromosome evaluation mechanism within the genetic algorithm framework，optimizing search strategies by dynamically balancing local path scoring and global contribution.Additionally，the algorithm introduces an adaptive crossover and mutation mechanism controlled by temperature coefficients，achieving a dynamic balance between exploration and exploitation.At the order processing level， this research constructs a "coarse-fine combined" dual-layer spatial clustering strategy，integrating dynamic grid clustering with improved K-means algorithm ， significantly reducing the computational complexity of large-scale order processing.Simulation results demonstrate that the algorithm achieves an order acceptance rate of 84.25% during peak hours，while maintaining average response waiting time and average riding time at 1.54 minutes and 3.59 minutes respectively，showing significant advantages over existing algorithms.This research provides a new technical approach to enhance the operational efficiency of DRT systems.